SCLin's neuroscience blog

Dopamine neurons don't just encode valence -- Some encode salience

2009-05-31T14:37:00.009-04:00

Two types of dopamine neuron distinctly convey positive and negative motivational signals

Nature advance online publication 17 May 2009.

Masayuki Matsumoto & Okihide Hikosaka

This is the classical kind of (putative) DA neurons, which show opposite responses to reward and punishment, thus encoding the valence of a stimulus.

This classical picture is well and good. But not so fast...

Mastumoto and Hikosaka here show that there is in fact another group of (putative) DA neurons, which have similar bursting responses to both reward and punishment. Thus, it appears that these DA neurons are encoding the salience of a stimulus, reminiscent of how basal forebrain neurons encode salience.

Hippocampal theta oscillations are travelling waves

2009-05-31T14:22:00.014-04:00

Evgueniy V. Lubenov & Athanassios G. Siapas

Nature 459, 534-539 (28 May 2009)

A movie is worth a thousand words. This is cool!

Define anatomic boundary by large scale expression pattern analysis

2009-01-19T11:10:00.003-05:00

Genomic Anatomy of the Hippocampus
Thompson et al, Volume 60, Issue 6, 26 December 2008, Pages 1010-1021 -- Neuron

What features really define the boundaries of neural systems? Anatomical landmarks and morphology have long been the guide, but these criteria come with obvious caveats that, at the transition zone where one region morphs into another, the determination of the exact boundary may be somewhat arbitrary. Is there an abrupt transition? or graded transition? and are there hidden subdivisions?

The authors here provides novel insight on this issue using genome scale gene expression data. Thompson et al applied powerful statistical pattern recognition tools to determine functional domains within the hippocampus based on the spatial expression pattern of ~3000 genes. This analysis not only recovers the major divisions of the hippocampus - DG, CA3, CA1 - but also uncovers subdivisions within. I really like this data-driven approach, which relies on the same tools that are used to uncover patterns in neurophysiological data.

Many molecules that define the map are adhesion molecules, which are important for forming neuronal circuits and finding projection targets. Supporting this idea, the subdivisions described in this paper (lower row) correspond to the spatial divisions in several previous reports studying the input-output relations of the hippocampus with other regions (upper row). Thus, this approach provides the molecular underpinning for these subdivisions and identifies molecular targets that can uniquely specify a functional domain. On a broader perspective, the same approach can be applied to all brain regions and shed new light on the functional organization of the brain.

Tracking of State Value in the Amygdala

2008-11-10T17:22:00.002-05:00

Moment-to-Moment Tracking of State Value in the Amygdala
Belova et al. 28 (40): 10023 -- Journal of Neuroscience

The amygdala is important for the regulation of emotion and also in learning. Neurons in the amygdala changes their response properties during learning. But what exactly are the amygdala neurons doing? what are they encoding? what is their computational role?

Previous studies have establish that amygdala neurons encode the motivational value of the associated outcome. When animals learn that a cue CS is associated with an outcome US, a subset of amygdala neurons show stronger responses to CSs that are paired with rewarding USs (positive value coding, example neuron A at 1-2 sec). Conversely, another subset of neurons code negative value and respond more vigorously to CSs paired with punishment (like airpuff, example neuron B at 2-3 sec).

The current study extends these previously findings to suggest that these amygdala neurons in fact code the value of the state in the generic sense. In essence, what the authors found was that neurons encode value not only for the CS, but also for other behavioral epochs, including the fixation point and also to the US. In the above examples, neuron A prefered rewarding CS and also showed stronger response to the fixation point, while neuron B preferred aversive CS and reduced its firing rate to the fixation point. This pattern is generally true for the population (see D).

I think this paper is really powerful because the conclusion is based not only on one behavioral epoch but on all the epochs. This comes much closer to understanding the roles of these amygdala neurons, which is to keep track of the current state value. Since neurons are active all the time and in various behavioral contexts, a thourough understanding of their physiological/behavioral roles require considering all these scenarios, not just any particular one.

This paper is also important because the coding of state value has important implications in computational theories of learning.

Neural correlate of decision confidence and uncertainty

2008-09-01T12:25:00.006-04:00

Neural correlates, computation and behavioural impact of decision confidence

Adam Kepecs1, Naoshige Uchida1,2, Hatim A. Zariwala1,3 & Zachary F. Mainen1,4

Humans and other animals must often make decisions on the basis of imperfect evidence^1,². Statisticians use measures such as P values to assign degrees of confidence to propositions, but little is known about how the brain computes confidence estimates about decisions. We explored this issue using behavioural analysis and neural recordings in rats in combination with computational modelling. Subjects were trained to perform an odour categorization task that allowed decision confidence to be manipulated by varying the distance of the test stimulus to the category boundary. To understand how confidence could be computed along with the choice itself, using standard models of decision-making^3,^4,^5,⁶, we defined a simple measure that quantified the quality of the evidence contributing to a particular decision. Here we show that the firing rates of many single neurons in the orbitofrontal cortex match closely to the predictions of confidence models and cannot be readily explained by alternative mechanisms, such as learning stimulus–outcome associations^7,^8,^9,¹⁰. Moreover, when tested using a delayed reward version of the task, we found that rats' willingness to wait for rewards increased with confidence, as predicted by the theoretical model. These results indicate that confidence estimates, previously suggested to require 'metacognition'^11,¹² and conscious awareness^13,¹⁴, are available even in the rodent brain, can be computed with relatively simple operations, and can drive adaptive behaviour. We suggest that confidence estimation may be a fundamental and ubiquitous component of decision-making.

This is a really nice paper by Adam Kepecs et al from CSHL. The gist of the paper is in the following figure. The model (a-d) illustrates the perceptual decision process of determining which category the odorant stimulus (s) belongs to, by comparing the stimulus (si) to the memory of the boundary (bi), each drawn from its respective distribution. This leads to the determination of the perceptual category (si>bi or si less than bi) and the decision confidence (absolute difference between si and bi). The behavior of the decision confidence construct (d) is matched perfectly with neuronal responses in the OFC (e, g).

One important concern of the authors' interpretation is whether the OFC neuronal activity signals uncertainty or an error signal. This is nicely addressed in the method section:

The observed selectivity of neural activity for the upcoming outcome might arise if, after executing a choice, extra sensory or memory information enters decision-making circuits and causes the realization that an error occurred even before obtaining feedback. According to this interpretation the negative outcome selective population of OFC neurons would signal error⁴⁴ instead of uncertainty. In contrast, the highest observed firing rates were associated with near chance level performance and not errors (Fig. 4g, f). To test this more rigorously, we asked whether an ideal observer could obtain better performance than the experimental subject if it could switch choices based on the firing rate after the choice and before feedback is provided. In all but one negative outcome selective neuron (1/133), the highest firing rates (top 5% of trials) were associated with chance level performance (within the 95% confidence interval). Therefore negative outcome selectivity does not imply that OFC neurons are actually able to predict error trials but rather that high firing rates predict near chance level performance consistent with an uncertainty signal.

The discovery of this neural correlate of decision confidence opens many new research questions. How is uncertainty/confidence calculated in the neural circuits? What are the influences of decision confidence on behavioral responses, learning and memory, and attention? No doubt these questions will be further explored by combining elegant experimental designs with a sound theoretical framework. Excellent work!

Synaptic plasticity differs for D1- and D2-striatal neurons

2008-09-01T10:48:00.002-04:00

Neurons in the same region are not created equal. The exact cell type matters, and matters a lot.

Dichotomous Dopaminergic Control of Striatal Synaptic Plasticity
Weixing Shen,1 Marc Flajolet,2 Paul Greengard,2 D. James Surmeier1*

At synapses between cortical pyramidal neurons and principal striatal medium spiny neurons (MSNs), postsynaptic D1 and D2 dopamine (DA) receptors are postulated to be necessary for the induction of long-term potentiation and depression, respectively—forms of plasticity thought to underlie associative learning. Because these receptors are restricted to two distinct MSN populations, this postulate demands that synaptic plasticity be unidirectional in each cell type. Using brain slices from DA receptor transgenic mice, we show that this is not the case. Rather, DA plays complementary roles in these two types of MSN to ensure that synaptic plasticity is bidirectional and Hebbian. In models of Parkinson's disease, this system is thrown out of balance, leading to unidirectional changes in plasticity that could underlie network pathology and symptoms."

What are the functional consequences of the differential regulatory rules for synaptic plasticity in D1- and D2-striatal neurons?

In the absence of behaviorally important stimuli, DA neuronsspike autonomously to maintain striatal DA concentrations atlevels sufficient to keep high-affinity D2 DA receptors active,but not low affinity D1 DA receptors —in principleenabling bidirectional, Hebbian plasticity in D2 MSNs, but notin D1 MSNs, where the low level of D1 receptor activity shouldpermit only LTD. However, when behaviorally important stimulidrive phasic spiking of mesencephalic DA neurons, striatal DAlevels rise transiently and activate D1 DA receptors ; thisshould enable the induction of Hebbian LTP in D1 MSNs.

Rubber hand illusion

2008-07-18T23:36:00.003-04:00

THE INVISIBLE HAND

"In the 'rubber hand' illusion, a person's hand and an adjacent rubber hand are both brushed gently. The real hand is kept out of sight. Before long, the subject's brain creates a new spatial link, imagining that the sensation in the real hand is arising where the rubber hand is."

Graduate student Matthew Mulvey of Leeds Metropolitan University has now shown that the effect will work if the researchers deliver transcutaneous electrical nerve stimulation (TENS) not to the hidden hand but to the wrist. After being primed with the illusion, subjects perceive the impulses--which hijack the nerve pathways between hand and brain--as a tingling located in the rubber hand.

This is such a cool idea to take advantage of this "rubber hand" illusion as a way of establishing sensory feedback on a neuro-prosthetic device.

Neuronal Ensemble Bursting in the Basal Forebrain Encodes Salience Irrespective of Valence

2008-07-09T13:54:00.004-04:00

Neuronal Ensemble Bursting in the Basal Forebrain Encodes Salience Irrespective of Valence
Shih-Chieh Lin and Miguel A.L. Nicolelis

Our paper finally comes out in Neuron today, accompanied by a preview from Lau and Salzman:

Although noncholinergic neurons in the basal forebrain are known to contribute to cognition, their response properties in behaving animals is unclear. In this issue of Neuron, Lin and Nicolelis demonstrate that these neurons represent the motivational salience of sensory stimuli and may modulate cortical processing to direct top-down attention.

This is our abstract and the main figure

Both reward- and punishment-related stimuli are motivationally salient and attract the attention of animals. However, it remains unclear how motivational salience is processed in the brain. Here, we show that both reward- and punishment-predicting stimuli elicited robust bursting of many noncholinergic basal forebrain (BF) neurons in behaving rats. The same BF neurons also responded with similar bursting to primary reinforcement of both valences. Reinforcement responses were modulated by expectation, with surprising reinforcement eliciting stronger BF bursting. We further demonstrate that BF burst firing predicted successful detection of near-threshold stimuli. Together, our results point to the existence of a salience-encoding system independent of stimulus valence. We propose that the encoding of motivational salience by ensemble bursting of noncholinergic BF neurons may improve behavioral performance by affecting the activity of widespread cortical circuits and therefore represents a novel candidate mechanism for top-down attention.

Reward-encoding in dorsal raphe (serotonergic) neurons

2008-05-17T09:18:00.003-04:00

Reward-dependent modulation of neuronal activity in the primate dorsal raphe nucleus.
J Neurosci. 2008 May 14;28(20):5331-43
Authors: Nakamura K, Matsumoto M, Hikosaka O

Serotonergic neurons in the dorsal raphe (DR) constitute one of the major neuromodulatory systems. What is the role of DR neurons in encoding reward, in comparison with midbrain dopaminergic neurons? The authors noted several differences: (1) DA neurons encode reward prediction error -- responding to reward only when the reward size is larger or smaller than expected. DR neurons respond to both reward and reward-predicting cues, whether or not they are expected. (2) DA neurons respond to reward with a phasic bursting response, while DR neurons show slower tonic responses.
In addition, DR neurons are heterogeneous, some prefer large while others prefer small reward. Without an independent means of verifying the neurochemical identity, it remains unclear which subset of DR neurons are serotonergic neurons.

Gamma response in EEG caused by micro-saccade

2008-05-09T20:44:00.003-04:00

Transient Induced Gamma-Band Response in EEG as a Manifestation of Miniature Saccades
Shlomit Yuval-Greenberg, Orr Tomer, Alon S. Keren, Israel Nelken and Leon Y. Deouell

Turns out that a large component of gamma oscillation responses in the EEG literature is generated by small eye movements. Caution: this is not to discount the physiological importance of gamma oscillation and its roles in cognition, as had been demonstrated in single unit studies, local field potentials, MEGs, and some EEG studies.

Heterogeneity of midbrain dopamine neurons

2008-04-27T10:57:00.001-04:00

Unique properties of mesoprefrontal neurons within a dual mesocorticolimbic dopamine system
Lammel S, Hetzel A, Hackel O, Jones I, Liss B, Roeper J

Here, we report the identification of a type of dopaminergic neuron within the mesocorticolimbic dopamine system with unconventional fast-firing properties and small DAT/TH mRNA expression ratios that selectively projects to prefrontal cortex and nucleus accumbens core and medial shell as well as to basolateral amygdala. In contrast, well-described conventional slow-firing dopamine midbrain neurons only project to the lateral shell of the nucleus accumbens and the dorsolateral striatum. Among this dual dopamine midbrain system defined in this study by converging anatomical, electrophysiological, and molecular properties, mesoprefrontal dopaminergic neurons are unique, as only they do not possess functional somatodendritic Girk2-coupled dopamine D2 autoreceptors.

I am curious to know whether mesoprefrontal DA neurons are not inhibited by apomorphine, which has been the criteria for identifying DA neurons recorded in vivo.

Neuronal oscillations and Attentional selection

2008-04-27T08:52:00.002-04:00

Entrainment of neuronal oscillations as a mechanism of attentional selection
Lakatos P, Karmos G, Mehta AD, Ulbert I, and Schroeder CE

The authors measured V1 CSD in monkeys performing an inter-modal attention task. In this task, visual and auditory stimuli were presented alternately. The task was to detect the odd ball stimulus in either the visual domain or in the auditory domain (in separate blocks). The main finding is panel A, where they showed that in layer II/III an opposite pattern of CSD profile in attend-visual condition vs. attend-auditory condition prior to stimulus onset. This is in contrast to the similar CSD patterns in layer IV-VI in both conditions. This difference reflected a layer-specific task/attention effect.
This attentional effect was interpreted as the brain exploiting the rhythmicity of stimulus presentation and recruiting a cortical oscillation at the same temporal frequency as a way to enhance stimulus processing. Alternative, the authors acknowledged that these results "may represent spontaneously occurring neural oscillations aligned by phase resetting to the structure of attended stimulus streams". In other words, it is unclear whether the stimulus resets the ongoing oscillation, or a top-down attention mechanism generates an oscillation.

Decoupling through Synchrony

2008-04-27T08:29:00.001-04:00

Decoupling through Synchrony in Neuronal Circuits with Propagation Delays
Lubenov EV & Siapas AG

This paper builds upon a powerful, yet simple, insight: that is, when two connected neurons fire at the same time (as in population bursting), the conduction delay would decrease their synaptic weight under the Hebbian STDP rule (A & B). Thus, population synchrony leads to a paradoxical weakening of synaptic connections. The consequence of this observation is that it "promotes the self-organization of spontaneously active neuronal network to a state at the border between randomness and synchrony". Applied to the synchronous activity of hippocampal network during slow-wave sleep, the authors suggest that this might provide a mechanism of selective erasure of memory traces in the hippocampus as memories are transferred to the cortex.

Neuronal Ensemble Bursting in the Basal Forebrain Encodes Salience Irrespective of Valence

2008-02-25T15:28:00.003-05:00

This is the abstract of the talk I will present this week at COSYNE meeting.

The main goal of animal behavior is to maximize reward and avoid punishment. To achieve this goal, animals similarly attend to both types of motivationally salient events despite their opposite hedonic valence. Recent evidence suggests that the opposite hedonic valence of reward and punishment are processed by different and possibly opposing neural systems. However, it remains unknown whether motivational saliency of reward and punishment is processed by the same valence-specific neural systems, or alternatively, is encoded separately as a distinct and valid neurobiological construct.

Here we show that motivational saliency is encoded by ensemble bursting of basal forebrain (BF) neurons in behaving rats. We observed that motivationally salient sensory cues that predicted either sucrose or quinine delivery in a Go/Nogo task elicited a similar brief bursting response in many BF neurons. This bursting response occurred irrespective of the cue’s sensory modality, the associated motor response and the hedonic valence of the expected outcome. BF ensemble bursting emerged as cues acquired motivational saliency and predictive ability through associative learning and diminished after the cue-outcome associations underwent extinction. The same BF neurons also responded to both primary reward (sucrose) and punishment (quinine) with highly similar bursting patterns. These salience-encoding BF neurons represented a homogeneous subset of BF non-cholinergic neurons because they did not change their average firing rate across wake-sleep states [1] and their firing properties were consistent with in vitro characterizations of BF non-cholinergic neurons [2]. Finally, the relationship between BF ensemble bursting and behavioral performance was documented by showing that BF bursting responses predicted successful detection of near-threshold tones in a tone detection task.

Our results point to the existence of an independent salience-encoding system, mediated by ensemble bursting of BF neurons. This discovery suggests that the valence and salience of attended stimuli are encoded by two major neuromodulatory systems – the midbrain dopaminergic neurons and BF non-cholinergic neurons – using similar bursting responses. Contrary to the traditional view that BF functions are mediated mostly via cholinergic neurons, our findings provide the first evidence regarding the in vivo functions of the poorly understood BF non-cholinergic neurons in behavioral contexts. The encoding of motivational saliency by BF ensemble bursting may improve behavioral performance by transiently enhancing cortical gamma oscillation power [3] and, therefore, mediating the influences of attention on cortical processing. Together, our results support the hypothesis that BF ensemble bursting represents a novel candidate mechanism for attention.

[1] Lee, M.G., et al., Cholinergic Basal Forebrain Neurons Burst with Theta during Waking and Paradoxical Sleep. J. Neurosci., 2005. 25(17): p. 4365-4369.

[2] Alonso, A., et al., Differential oscillatory properties of cholinergic and noncholinergic nucleus basalis neurons in guinea pig brain slice. Eur J Neurosci, 1996. 8(1): p. 169-82.

[3] Lin, S.C., D. Gervasoni, and M.A. Nicolelis, Fast modulation of prefrontal cortex activity by basal forebrain noncholinergic neuronal ensembles. J Neurophysiol, 2006. 96(6): p. 3209-19.

Non-cholinergic basal forebrain projections to prefrontal cortex

2008-02-21T19:08:00.003-05:00

Basal forebrain (BF), also referred to as nucleus basalis, is one of the largest neuromodulatory systems, which plays important roles in attention, arousal and the control of cortical activity and plasticity. Traditionally, these BF functions have been attributed to BF cholinergic projections to the cerebral cortex.

However, what is little known to the general neuroscience community is that BF provides more non-cholinergic projections to the cortex, mostly GABAergic neurons and some glutamatergic neurons. In a recent article by Henny P and Jones BE, the non-ACh BF projection is further investigated.

Eur J Neurosci: Projections from basal forebrain to prefrontal cortex comprise cholinergic, GABAergic and glutamatergic inputs to pyradimal cells or interneurons

This study establishes that BF projections to the prefrontal cortex consist of distinct (non-overlapping) ACh, GABAergic and glutamatergic systems, with the GABAergic system being the most prominent system (account for 52% of BF terminals). These three systems innervate all cortical layers and most prominently the deep layers (V-VI). All three systems innervate calbindin+ interneurons. In addition, BF GABAergic projections innvervation parvalbumin+ interneurons (basket and chandelier cells, which provide inhibition to the soma, promimal dendrites or axon initial segment of pyramidal cells).

What are the functions of these non-ACh BF projections? In a previous paper, I demonstrated that BF non-ACh neurons spontaneously engage in ensemble bursting. Most importantly, I showed that BF ensemble bursting may lead to transient enhancement of prefrontal cortex activity (e.g. increase gamma oscillation). This finding is consistent with BF GABAergic projections to the PV+ interneurons, which provide the anatomical substrate to transiently enhance cortical activity.

What are the behavioral functions of BF ensemble bursting? This is the topic of my current study, which I will present this week at the COSYNE meeting. More on that later.

Neuroscience goes global

2008-01-21T22:18:00.000-05:00

Neuroscience is growing fast in mainland China, as reported it Nature Neuroscience Editorial.

The Institute of Neuroscience in Shanghai, founded in 1999 and led by Berkeley's Mu-ming Poo, has inspired reform of existing institutions and the establishment of newer ones. The National Institute of Biological Sciences, opened in 2004 on the outskirts of Beijing and led from afar by Xiaodong Wang of the University of Texas Southwestern, is modeled on the Howard Hughes Medical Institute. It provides generous support to promising scientists without prescribing projects or approaches...

These new neuroscience institutes in China have been able to recruit top-notch scientists by offering competitive startups rivaling those offered in the US. These institutes thus provide enviable research environments not only for the generous funding level but, most importantly, for creating the critical research mass that is essential for producing competitive science. And they are doing really well.

Similar neuroscience institutes aiming for international excellence are also found in Japan, featuring Riken Brain Science Institute and the recently established Okinawa Institute of Science and Technology; In Portugal, there is the new Champalimaud Foundation Neuroscience Programme; In Brazil, my adviser Dr. Nicolelis is building a neuroscience institute at Natal, as part of his vision to promote science and education.

As a national of Taiwan, we are way behind in this wave. It is my sincerest hope that similar effort can find support in Taiwan. These recent developments point to the globalization of neuroscience research. These examples not only prove that building a world-class neuroscience institute can be done, it is urgent that we start now.

Ed Boyden's blog

2007-12-16T23:45:00.000-05:00

I recently came across Ed Boyden's blog on Technology Review and I thought it's worth sharing. Ed Boyden is currently a faculty at MIT media lab. I first knew of his work from researching recent developments of new optical techniques that can control neuronal activity. His recent blog entry on How to think caught my attention, which I agree with for the most part but I am not able to do it all. Here's an excerpt:

1. Synthesize new ideas constantly. Never read passively. Annotate, model, think, and synthesize while you read, even when you're reading what you conceive to be introductory stuff. That way, you will always aim towards understanding things at a resolution fine enough for you to be creative.
From: Boyden, E. S. "How to Think." Ed Boyden's Blog. Technology Review. 11/13/07. (http://www.technologyreview.com/blog/boyden/21925/).

And he's only 27 now. Wow! That's impressive!

Some useful tips on...

2007-11-24T20:45:00.000-05:00

It takes so much time and effort to complete a set of experiments and to write up a paper. After finishing the manuscript of my first paper, I thought the end was near. Of course, it was far from over. Besides the manuscript itself, there are several other components involved in getting the paper published. Here are some articles that I came across that should be helpful.

Joshua Finkelstein (Senior Editor of Nature) offers some advices on how to write a cover letter to the editor
How to review a paper and some tips from an editor
How to read rejection letters and tips on how to respond to reviewers' criticism.

Some random thoughts about this blog

2007-11-21T13:25:00.001-05:00

Obviously, I have not been posting for quite a while. Life as a postdoc has been more crazy than I imagined, and I am sure it will only get worse over time.

At SfN this year, I am pleasantly surprised by a few conversations where people actually read this blog. Thank you if you are one of those people. On the other hand, not maintaining a steady flow of post is almost like not mowing the lawn. I do feel guilty about it and I will try to keep it up.

Another interesting thing I learned is that some people thought I might be the reviewer of their paper, probably because I commented on their paper on this blog in ways similar to their reviewers. I just want to clarify that this is absolutely not the case. I post good papers that I find interesting, and write comments as a way to think more deeply about the results. That's all.

Happy Thanksgiving.

How cholinergic modulation changes cortical circuits

2007-11-21T10:14:00.000-05:00

Nature: A synaptic memory trace for cortical receptive field plasticity
Froemke RC, Merzenich MM, Schreiner CE

This is an excellent paper revealing how cholinergic modulation (by stimulating nucleus basalis, or basal forebrain) transiently changes the balance between excitation and inhibition within the cortical circuit, thus mediating specific circuit plasticity in the primary auditory cortex. In pentobarbital-anesthetized rats, the authors showed that pairing tone presentations with NB stimulations leads to immediate suppression of inhibition (<20s) followed by enhancement of excitation (start around 40s), specifically for inputs tuned at the paired frequency and blocked by muscarinic antagonist atropine. This unbalanced excitation and inhibition takes about 1-2 hours to return to baseline (provided continual auditory stimulation), thus providing a window for synaptic plasticity and modification of receptive field. Thus, NB stimulation leaves a tag of reduced inhibition (at the circuit level!), specifically for neurons that are active at the time of pairing.

The paper devotes figure 3 to address whether NB pairing specifically modifies intracortical connections vs. thalamocortical projections, and showed that only intracortical synaptic strengths are modified while thalamocortical transmission unchanged. I believe the unspoken message here is the contradicting results previously obtained in slices (by Hasselmo) that show ACh suppression of intracortical transmission and enhancement of thalamocortical transmission.

What accounts for the difference? The main difference is the preparation. Froemke et al studied cortex in vivo in anesthetized rats and electrically stimulated NB. The NB neuromodulatory systems contain both ACh and non-ACh projections that might be both important for controlling cortical activity and plasticity. More importantly, ACh release following NB stimulation better mimics the spatio-temporal pattern of physiological ACh modulation, unlike bath application of ACh agonist onto the slice. Still, the detailed mechanisms leading to opposite observations in two preparations will need to be worked out.

This is a set of technically challenging experiments. The main question I hope the authors would address in the future is regarding the temporal relationship between tone presentation and NB stimulation. In the current report, NB stimulation seems to start at the same time with tone presentation (but I cannot find detailed description of this procedure, nor the number of tone presentations and number of pairings involved). This is unlikely to be true in physiological conditions, as the activation of NB neurons naturally would lag 50-100 msec after tone onset. By the time ACh modulation reaches cortex, the neuronal activation traces induced by the 50 msec tone might have dissipated. Would NB stimulation remain effective in that case? How many pairings are required to generate a lasting effect? This should be a sufficiently small number to effective in normal learning.

Addendum: Following the thought on looking at the temporal relationship between tone presentation and NB stimulation, determining the relevant time window is important because a broad window would mean any sensory stimulus present during this period will be potentiated. Since many potential cues exist in the environment, a broad temporal window for effective NB stimulation will significantly reduce the specificity of this mechanism. This issues can be tested without intracellular recording though.

OFC and outcome encoding

2007-05-09T17:47:00.000-04:00

J Neurosci: Orbitofrontal cortex mediates outcome encoding in pavlovian but not instrumental conditioning.
Ostlund SB, Balleine BW.

I've always had the impression that in the neuroscience community among people studying mammalian behavior, the rich and solid learning theory and behavioral tradition from the psychology root have not been well appreciated as they should. We all know, or think we know, about pavlovian and instrumental conditioning. But subtle variations in these 'simple' behaviors are not so simple at all. In recent years, great stride has been made about the neural systems subserving various 'constructs' in learning theories. The circuit-level understanding of behavior is rapidly evolving. The use of rodent to dissect circuitry proved to be a powerful tool, especially in the hands of great learning theorists, such as Bernard Balleine here and also Peter Holland. As a neuroscientist studying rodent neurophysiology, I often marvel at the insights their experiments provides - usually involving no more than behavior and lesion. I highly recommend their works.

This is another solid paper from Balleine. It shows that lesions of orbitofrontal cortex impairs outcome encoding, but only in the context of pavlovian but not instrumental conditioning. In other words, the outcome representation in stimulus-outcome association (pavlovian) is mediated by OFC; while the outcome representation in action-outcome association (instrumental) is mediated by prelimbic cortex (Balleine and Dickinson 1998, see discussion in the paper).

How do they reach this conclusion? Rats are trained to learn Cue1 --> Reward1, Cue2 --> Reward2 ; then learn Action 1--> Reward1, Action 2--> Reward2; Before the testing phase, give rats free access to Reward1 until satiation, so that the value of Reward1 decreases. Under this condition, the decreased value or Reward1 is reflected in rats' tendency to perform Action 2 more often than Action 1. This action-outcome association is not impaired by OFC lesion.

However, if during the testing phase, Cue1 and Cue2 are presented, Cue1 would normally potentiate Action1, through stimulus-outcome association. This process is called outcome-specific pavlovian-instrumental transfer. OFC lesion impairs this process.

In the second experiment, train rats to learn Cue1 --> Reward1, Cue2 --> Reward2; and subsequently degrade the relationship between Cue1 and Reward1. Normal rats would selectively decrease their responses toward Cue1, but OFC lesion rats would be impaired toward both cues.

Lesion studies are great, especially in showing double dissociations, in this case the dissociation between OFC and PL in outcome encoding in two different contexts. One caution regarding the lesion approach is that brain is a dynamic structure; information flows between areas and through this combination of feed-forward and feedback processes some conclusions are reached. To translate lesion works into understanding the actual dynamics of the system is no simple task, but they are a great start.

How new imaging techniques are changing neuroscience

2007-04-09T11:20:00.000-04:00

In case you haven't caught up with it, exciting new techniques are being developed in the imaging community. These techniques can, and are already changing the landscape of neuroscience.

Several recent papers come up with intriguing tricks to control neuronal activity with LIGHT at millisecond temporal resolution (a review). The idea is to introduce light-driven ion channels from algae into mammalian cells, which turn lights into either excitatory (channelrhodopsin-2, see this and this) or inhibitory currents (chloride channels, see this and this) . These channels appear to be non-toxic, non-adapting to prolonged stimulation, and respond to light stimulation with millisecond temporal resolution. Couple these new tools with techniques that target gene expression in selected brain region or even cell type (using virus vectors or lox/cre system), then you can turn on or off your favorite neuron by shining a light.

What can you do with this? Study the causality! The formula goes like: Does the activity of neuron type X at certain time point T is required for behavior Y. Turn it on or off at your choice. Electrical stimulation is of course no comparison. The only limitation seems to be genetic manipulations. But just like tons of knock-out mice now available, it is only a matter of time before this becomes the standard practice. Physiologists be aware!

Sleep consolidates emotional memory

2007-03-15T12:31:00.000-04:00

See Commentary in Current Biology

What does sleep has to do with memory? Many recent studies have established that sleep consolidates some types of memory, most notably procedural memory such as learning to ski or bike. However, it remains an settled issue whether sleep enhances episodic memory. This commentary nice summarized findings of two recent papers, both converging to the conclusion that sleep selectively enhances episodic memory that is emotionally salient, but not emotionally neutral words/pictures. This commentary also provides some nice references on the role of sleep in memory consolidation.

The speculations in this commentary is quite intriguing. If sleep enhances emotional memory, is it a natural protection mechanism to be insomniac when we are anxious or depressed, thus avoiding the impact of negative emotions? Does antidepressant drugs achieve their roles partly by inhibiting REM sleep? More provocatively perhaps, what about "sexual activity has a known soporific effect" and may facilitate bonding of romantic partners?

How do you keep up with the literature?

2007-03-01T21:42:00.000-05:00

Information is exploding. Just reading the table of contents from the major journals plus some interesting abstracts would take some time, not to mention the discipline needed for such habit. And yet, I was always worried that I missed out on some interesting/relevant paper. What's the solution?

Well, I have tried subscribing to journal TOCs and getting my pubmed search results sent to me weekly (set up an My NCBI account in pubmed, it's really useful). But pretty soon these email alerts were swamped in my gmail inbox, and I have many other emails to deal with. So either I ended up forget to go back to those email alerts, or that going through individual emails and find out the right piece of paper was just too time-consuming.

I wondered, if there's a simple interface where all the relevant information is at my fingertip, and I can just browse through all the relevant info (title/abstract/author) whenever I feel like it, without fearing that I lost track of something. Here's solution I started using lately, which I found quite satisfactory: using Google Reader to track pubmed RSS feed. With this trick, you can keep up with your pubmed search results through RSS feed. So whenever ANY new paper matching your search criteria comes online, it would appear in your Google Reader. You can also subscribe RSS feed of major journal TOCs from journal websites, from this website, or through pubmed. So here you go, a central deposit of all the latest journals and papers of interest only one click away (and no ads).

How do you deal with the vast amount of literature? Share your thoughts.

Random sample

2007-02-09T17:01:00.000-05:00

PNAS: Corticothalamic feedback enhances stimulus response precision in the visual system
Ian M. Andolina, Helen E. Jones, Wei Wang, and Adam M. Sillito
Why is there massive and more prominent corticothalamic feedback projection from V1 to LGN? By reversibly inactivating V1 with muscimol while simultaneously recording from LGN neurons, this paper demonstrates that LGN response timing and precision are significantly improved when cortical feedback is intact.

PNAS: Sequential structure of neocortical spontaneous activity in vivo
Artur Luczak, Peter Barthó, Stephan L. Marguet, >György Buzsáki, and Kenneth D. Harris
How does local cortical circuit organize its activity pattern? By studying the activity of many cortical neurons during DOWN-UP state transitions in anesthetized and awake rats, the authors observed stereotypical activation sequence of local cortical neurons during this transition, the pattern is especially precise during the first 100 msec. This sequence also seems to give rise to repeating motifs of population spike patterns (like cortical songs from Yuste lab). It would be really interesting to know if learning changes this activation sequence (and hence local circuitry)!

PNAS: Patients with hippocampal amnesia cannot imagine new experiences
Demis Hassabis, Dharshan Kumaran, Seralynne D. Vann, and Eleanor A. Maguire
This is quite interesting! Hippocampus is involved not only in recalling past episodic memory but also in constructing imagination of new experiences! So how much of memory is a construction process? And what does hippocampus really do? Also read comments in Science.

Another random sample

2007-02-08T15:48:00.000-05:00

Neuron: Neuronal oscillations and multisensory interaction in primary auditory cortex
Lakatos, P. - Chen, C. M. - O'Connell, M. N. - Mills, A. - Schroeder, C. E.
I heard Dr. Schroeder present this data some time ago and was quite impressed. The findings point to a novel mechanism in the multisensory interaction: the non-specific thalamic pathway seem to provide a reseting signal to all cortical areas, thus optimizing the timing of information processing. This would make a great journal club paper.

PLoS Biology: Timing and Sequence of Brain Activity in Top-Down Control of Visual-Spatial Attention
Grent-'t-Jong, T. - Woldorff, M. G.
Through the combination of ERP, fMRI and source localization of a same attention-orientation task, the authors dissect the timing and sequence of cortical activity. The results suggest that orientation initiated from the frontal cortex and only later involved visual cortex.

J Neurosci: Noncholinergic lesions of the medial septum impair sequential learning of different spatial locations
Dwyer, T. A. - Servatius, R. J. - Pang, K. C.
By comparing selective immunotoxic cholinergic lesion with kainic acid lesion (which preferentially destroys GABAergic neurons) in the medial septum, the authors demonstrated that KA lesion, but not cholinergic lesion, leads to slower learning of a hippocampal-dependent task. Taking this result to the parallel system between basal forebrain and the cortex, these findings would lend support to the functional importance of BF non-cholinergic projections (mostly GABAergic) to the cortex.

Random samples

2007-02-08T13:36:00.000-05:00

Nature: Categorization of behavioural sequences in the prefrontal cortex
Keisetsu Shima, Masaki Isoda, Hajime Mushiake and Jun Tanji
Activities of prefrontal cortex neurons encode the "category" of planned motor sequence. The example in Fig2 (during error trials) is quite striking. Together with earlier results from Miller lab, prefrontal neurons encode category information (high-order representation) of both sensory and motor representations. See also comments in NatNeurosci.

Nature: Open-access journal will publish first, judge later
PLoS One is the latest open-access journal from PLoS, which adopts a completely different model of publication from traditional journals. Each paper is reviewed by one editor to check sound scientific design and analysis, but not for significance and impact. The paper is then published first, and readers can comment of the paper online. At first glance, I find it hard to navigate through the papers, probably due to reading habit. The fundamental question is: given the vast amount of publications, how do we choose and decide which papers are good and worth-reading? The issue of peer-review and models of publication also deserve further discussion.

Nature Neuroscience:Separate neural substrates for skill learning and performance in the ventral and dorsal striatum
Hisham E Atallah, Dan Lopez-Paniagua, Jerry W Rudy & Randall C O'Reilly
Skill is learned first in the ventral striatum. Its performance depends on both ventral and dorsal striatum. Eventually the memory trace, once establishes as a habit, can be independent of the striatum.

pre-SMA and behavioral switching

2007-02-02T13:07:00.000-05:00

Nature Neuroscience: Switching from automatic to controlled action by monkey medial frontal cortex
Masaki Isoda & Okihide Hikosaka

How does the brain stops a habitual action and switches in favor of an alternative controlled action?

The experimental design is clever: two peripheral stimuli with different color (red and yellow) are presented as potential saccade targets in each trial, the correct target only revealed later by the matching-color of the central cue. The trick is that within a block, the cue remains the same color, thus creating a response tendency toward that target. Once the block switches and the central cue changes color, the pre-planed saccade has to be suppressed and a contralateral saccade be made. These switch trials are the focus of this paper.

The authors demonstrated that switch trials are associated with slower reaction time and higher error rate (Fig1). A population of neurons in the pre-SMA (supplementary motor area) is specifically activated by the switch trials (Fig2). More specifically, these neurons are active before saccade in correct switch trials; however on error switch trials, these neurons are active after saccade execution. Electrical stimulation of this area slows reaction time and enhances the percentage of correct responses (Fig 3). Overall, these data support pre-SMA neurons not only monitors the conflict in action selection but likely resolves the conflict by inhibiting the habitual action in favor of the controlled action. The interpretation is that in correct trials, pre-SMA neurons are active early enough to switch the action. In error trials, their activity came too late.

The authors proposed that this activity likely modulated action selection through its basal ganglia projection down to superior colliculus. But the question is, how general is this mechanism? Can it be applied to action selection not involving saccade?

Back in town

2007-02-02T11:42:00.000-05:00

I am back in town from my trip back to Taiwan. With the help of melatonin, I have (almost) recovered from the 12-hour jet lag. It's something I highly recommend trying, 1mg is more than enough.

Things are a bit hectic the first few days back to work. As I slowly catch with the literature over the past month, I will post some highlights of recent literature.

Goal-related activity in Hippocampal Place Cells

2007-01-18T09:10:00.000-05:00

Place Cells in rat hippocampus fire selectively at certain locations. These cells may therefore be said to represent locations in space. If a rat has learned a certain maze, then every location in the maze is associated with some neurons, and the rat's location can be reconstructed by the population firing pattern.

Anecdotally, place cell RFs (called "firing fields") are clustered near the goals of the maze (the platform in the water maze, the food in an arm maze). But data has been unclear. This paper provides clear data to show that this is true.

The rats are in a cylinder maze, and the goal is "hidden." That just means that a computer drops a pellet when the rat crosses a certain unmarked location. Once the rat learns the maze, many place cells acquire a secondary, weaker, response field at the goal location.

The authors suggest that these signals indicate the hippocampal cells represent monitoring of the rat's performance as well as providing a map of space. They do not speculate on how these unrelated signals are decoded, which would seem to be a major problem.

But I don't have any better explanations.

Happy New Year!

2006-12-31T13:11:00.000-05:00

Not much has been going on in this blog cause I have been quite busy lately. It will likely stay that way until I return by the end of January. I hope you all have a great year of the pig.

Top-Down Control of Motor Cortex Ensembles by Dorsomedial Prefrontal Cortex

2006-12-18T20:35:00.000-05:00

Neuron: Top-Down Control of Motor Cortex Ensembles by Dorsomedial Prefrontal Cortex
Nandakumar S. Narayanan and Mark Laubach

How does the dmPFC cortex exerts top-down modulation on motor cortex neuronal activity? (note: dmPFC in rats is analogous to PFC in primates)

The authors studied dmPFC and motor cortex (M1) neurons in a delayed-response task, where rats press and hold a lever for 1 sec and release after hearing a trigger auditory signal. Using this task, the authors demonstrated that (1) dmPFC neurons are significantly modulated during the delay period; (2) Inactivation of dmPFC increases premature response errors (early release); (3) Inactivation of dmPFC specifically decreased the proportion of delay-related modulation of M1 neurons; (4) dmPFC and M1 neurons are most strongly correlated during the delay period (6% total pairs). These findings provide a dynamic picture of neuronal ensemble interactions between dmPFC and M1, with strongest interaction during the delay period. Such interaction likely reflects a top-down modulation from dmPFC on M1.

The authors interpreted these findings strongly in favor of a "temporal processing" hypothesis, i.e. dmPFC is critical for the temporal control of behaviors and inhibit temporally inappropriate responses. The evidence supporting this hypothesis is not strong. Perhaps an easier alternative, as authors acknowledged, is that dmPFC serves the role of working memory to maintain the task requirement (keep pressing the lever) online.

Regardless of the behavioral functions involved, the temporal correlation between dmPFC and M1 indicates the two neuronal ensembles are interacting most strongly during the delay period. Since the percentage of neuronal pairs with significant correlation is low and their temporal patterns likely heterogeneous, the authors didn't provide further indication of how they interact. Since the authors also collected LFP data, it might be interesting to look at oscillatory coupling using the framework Pascal Fries proposed.

Boosting slow oscillations during sleep potentiates memory

2006-12-05T14:43:00.000-05:00

The notion that sleep is beneficial for memory consolidation or even "enhancement" has gained increasing support in the recent years. In human studies, different kinds of memory depend on different sleep stages -- declarative memories, such as word-pair association, often rely on slow-wave sleep (SWS), while non-declarative memories, such as sequential finger tapping or visual discrimination, rely on rapid-eye-movement (REM) sleep. By deprivation of sleep or certain sleep stage, people have shown the mnemonic function of sleep. However, the mechanism(s) that sleep contributes to memory are unclear, and are hard to study. It could be on the level of electrophysiology, i.e. change of network properties during sleep, and/or on the cellular/molecular level, changing synaptic weights, etc.
The current paper is one of the seminal studies that have tried to manipulate one component of sleep in order to enhance memory consolidation, and shed lights on the potential mechanism. In this study, the authors trained the subjects of paired-asscociate word learning task before sleep, and induced artificial slow-waves (0.75Hz, transcranial stimulation) just before the subjects fell into SWS in the first 45 min of sleep, and tested their memory on the subsequent morning. There is an enhancement after sleep for all subjects, but more importantly, the group that received the stimulation gained more than the sham group did. They checked one more declarative task and two more procedural tasks to confirm that such effect only exists for declarative tasks. They did some further control for non-specific stimulation effect, by stimulating in the last 45 min of sleep, and did not get the enhancement. They also tried another stimulation frequency (5Hz, theta band), and that stimulation did not have such effect either.
Next they looked into the EEG recordings and tried to understand what signiture was changed by SW stimulation. They found increase of the length of SWS, increase of power in slow wave and spindle frequency bands, and they speculated that it was the increase of spindle oscillations that contribute to the enhancement induced by SW stimulation.
I think this paper is a break-through in the field, because it is the first time that manipulating slow-waves enhances the memory consolidation (or slows down memory loss) during sleep. It reminds me of a 2003 EJN paper in which the authors tried 3 ways to increase the amount of REM sleep (two drugs and REM deprivation beforehand) and that enhanced the rats' retention performance. The current paper is even neater. I doubt whether their next paper is going to be theta stimulation in late night enhances retention of procedural learning tasks. Regarding the mechanism, they speculated that increase of spindles is the reason for such enhancement. I would think that increased length of SWS itself is also a factor, given that previous experiments showed an correlation between retention and length of SWS. In the accompanying comment from Robert Stickgold, he thought that acetylcholine (ACh) should be an important thing to look at, since the same group have shown that low ACh during sleep favors the consolidation of this kind of memory. I will tend to think that high ACh favors the REM-dependent memories on the other hand, but it is difficult to prove.

Predictive codes for forthcoming preception in the frontal cortex

2006-12-04T09:56:00.000-05:00

When asked to discriminate between two categoris of images, do people make decisions after enough evidence has been collected since the onset of a image (in the analogous way that monkey's MT does for deciding left or right movement), or, do people bare some caterogical template before image onset? The authors hypothesized that it is actually the second case and they set out to find brain regions that work for such purpose.
They did a pretty straight-forward fMRI experiment with hybrid of blocked and event-related design. Such design was quite elegant to address their question. The subjects were asked to discriminate between houses and faces. In each trial, they saw one image and decided which category the image belonged to. Brain regions revealed by these event-related activities are the ones that process the categorical images (such as fusiform area for processing faces, etc), which is quite known. However, in each block, the subjects were asked to only respond to either houses or faces, such that in a "face" block, they were expected to hold the concept of face, or mental "template" of faces in their brain. Then by comparing "face" vs "house" blocks, the authors looked for the regions that hold such templates. The regions they found were dorsal and ventral foci in medial frontal cortices (dMFC and vMFC). They further calculated a dynamic causal model for the MFC regions and actual processing regions, and interpreted the causal link as top-down influence from the MFC to the other regions to affect processing.
The question was quite important and the results are what I would like to see. The design was simple, but elegantly addressed their hypothesis. However, the evoked responses in MFC were conterintuitive to me: they dropped in both "face" and "house" block, although the drop was smaller in "face" block (thus larger than in "house" block). If these regions are holding some "template" during the task, why the evoked response would decrease upon each trial? And I am even not sure what the time 0 in the evoked responses in their figures. Should we look for responses earlier in MFC than the other regions (fMRI may not have such temporal resolution though)? Secondly, their causal model seemed to bare too many arbitrary causal assumptions which I don't like much. PS: the figure legends in Fig2B and Fig3B were mis-labeled. Labels for "H in H" and "H in F" should be switched (3rd and 4th should be switched).

Cannabinoids, spike timing and cell assembly

2006-11-30T14:10:00.000-05:00

Nature Neuroscience: Cannabinoids reveal importance of spike timing coordination in hippocampal function
David Robbe, Sean M Montgomery, Alexander Thome, Pavel E Rueda-Orozco, Bruce L McNaughton & György Buzsaki

How does cannabinoids, the primary psychoactive component of marijuana, impair episodic memory? The authors showed that intrahippocampal infusion of cannabinoids reduces oscillation power at theta, gamma and ripple range. This manipulation also impairs the performance of a delayed alteration memory task. The extent of memory impairment is correlated with the decrement of theta oscillation power (which can not be trivially explained by a change in running speed).

The intriguing finding of the paper is that, despite the lack of change in average firing rate at the time scale of >200 msec, the proportion of burst spikes significantly decreased. This effect impairs coordinated cell assembly firing patterns, and even reduces baseline spike correlation. The authors therefore concluded that cannabinoids impairs hippocampal-dependent memory by destroying proper temporal coordination among neurons.

How does this happen, i.e. not affecting firing rate but reducing temporal coordination among neurons? A likely cause is that cannabinoids acts to reduce both excitatory and inhibitory functions, which results in weaker cell assembly and synchronization. The consequence of disrupting fine temporal patterns (while preserving gross firing rate) is a major change in both the network dynamics and the overt behavioral consequences.

Given that CB1 receptor is also highly expressed in neocortex, cerebellum and the basal ganglia, I would expect similar effects to be found at least in the neocortex.

Random samples this week

2006-11-21T23:49:00.000-05:00

Nature: Influence of the thalamus on spatial visual processing in frontal cortex
Marc A. Sommer and Robert H. Wurtz
Mediodorsal thalamus provides the corollary discharge signal, an internal copy of one's own movement, to the frontal cortex.

PNAS: The spine neck filters membrane potentials
Roberto Araya, Jiang Jiang, Kenneth B. Eisenthal, and Rafael Yuste
Spine not only is a biochemical compartment, but also an electrical compartment. The paper also features a cool new way of measuring membrane potential locally (in the spine) using FM dye.

PNAS: Recent fear is resistant to extinction
Stephen Maren and Chun-hui Chang
Extinction training right after fear conditioning might have immediate effect, but does not reduce the fear memory in the long run. The level of anxiety at the time of extinction is the key. Congratulations to Chun-hui !

Directional Signals in the Prefrontal Cortex and in Area MT during a Working Memory for Visual Motion Task

2006-11-11T01:26:00.000-05:00

Daniel Zaksas and Tatiana Pasternak, University of Rochester
Journal of Neuroscience, November 8, 2006, 26(45):11726-11742

The authors studied neuronal responses in area MT and PFC while monkeys performed three different versions of motion direction detection/comparison task. The task involves both the sensory aspect (motion detection) and the decision aspect (comparison, delay). This allows the authors to study how the two areas participates in both aspects, and how they possibly interact.

The general findings are that MT neurons are more responsive and selective toward motion when compared to PFC neurons. MT neurons also have shorter response latencies. They compared neuronal responses during the sample period, delay period, test period, and also compared error trials and subjective choices. There are many interesting details. The one piece I especially like is that PFC neurons show sustained responses to the sample when it is task-relevant (panel C, upper), but only show transient responses when it's irrelevant (panel C, lower). This occurs despite the fact that MT neurons respond similarly (and possibly similar inputs into MT). What makes the difference then?

A major weakness of the paper is the fact that PFC and MT neurons are recorded in different monkeys. And monkeys with PFC implants also happen to perform worse. This behavioral difference weakens the finding that PFC neuronal responses are not as selective and occur later than MT neurons. Despite this point, I believe this result would still hold if recordings were done in the same animals.

Auditory Brainstem Timing Predicts Cerebral Asymmetry for Speech

2006-10-25T20:31:00.000-04:00

The Journal of Neuroscience, October 25, 2006, 26(43):11131-11137
Daniel A. Abrams,¹ Trent Nicol,¹ Steven G. Zecker,¹ and Nina Kraus¹^,2

A nice example of what kind of functional insight can be provided by the electrophysiology approach. The paper links a few ideas together: auditory brainstem response latency --> auditory cortical response asymmetry --> hemisphere asymmetry and dominance --> possible tie with dyslexia and language processing. What's amazing is that a small difference (0.5-1ms) in the auditory brainstem response is correlated with a big difference in cortical processing and behavioral performance.

A common weak point of the electrophysiology approach is that it can only reveal correlation, but not causality. To prove causality, in most cases, is very difficult. Microstimulation has its issues; functional inactivations, via pharmacological means or via TMS, are not pretty either. I think it suggests that electrophysiology is phenomenological, but not mechanistic.

Developing Intelligence

2006-09-22T20:56:00.000-04:00

Just came across a great neuroscience blog, Developing Intelligence, with a special focus on attention. It's really worth reading.

High Gamma Power Is Phase-Locked to Theta Oscillations in Human Neocortex -- Canolty et al. 313 (5793): 1626 -- Science

2006-09-19T10:15:00.000-04:00

This paper, using subdural recordings in human subjects, provides clear evidence that (1) theta oscillations are coupled with high gamma (80-150 Hz) oscillations, which occur preferentially at theta oscillation troughs (Fig 1B, 2A); (2) this coupling is specifically between theta and high gamma, and not between other frequencies (Fig 1D); (3) these couplings are local, since they do not appear in all electrodes (Fig 3B), (4) The spatial extent/distribution of theta-gamma coupling is indicative of the cognitive task the subjects were performing, being more similarly distributed when subjects performed similar tasks (Fig 3D,E). These results therefore "support the hypothesisthat cross-frequency coupling between distinct brain rhythmsfacilitates the transient coordination of cortical areas requiredfor adaptive behavior in humans".

The distinction between high vs. low gamma oscillation is quite interesting. It appears that high and low gamma oscillations are dissociated here, with the high gamma more similar to the hippocampal ripple oscillation in rats. They have a recent paper dealing with this issue in depth.

How did this coupling take place? The authors referenced to my recent paper and suggested that basal forebrain non-cholinergic neurons could play a role in the coupling. It's great for me to know that the phenomena I observed in rats also exist in human.

How Visual Stimuli Activate Dopaminergic Neurons at Short Latency -- Dommett et al. 307 (5714): 1476 -- Science

2006-09-11T17:22:00.000-04:00

Midbrain dopaminergic neurons are generally considered to encode rewards. However, they respond to a wider range of salient stimuli, not necessarily associated with rewards. This paper is trying to address such responses wth electrophysiology (mainly extracellular unit recording) and electrochemistry (amperometric recording of extracellular dopamine level). They knew that deep layers of superior colliculus(SC) "project directly" to DA neurons, and that whole-field light flash will induce colliculus responses. Under anesthesia, DA neurons do not respond to the flash, except when GABA antagonist is applied to SC to disinhibit the SC. So they recorded from both SC neurons and midbrain DA neurons, and showed "short-latency" responses in both sites: ~40ms in SC and ~110ms in DA neurons. Unlike the reward-associated role of DA neurons which habituate after repeated stimuli presentation without rewarding, the responses in the current paper were demonstrated to sustain over hundreds of trials. They also measured striatal DA concentration for the same stimuli, and showed transient increases locked to stimuli. To summarize, they showed DA responses to salient visual stimuli with electrophysiology and electrochemistry.One thing I was wondering is that if those DA neurons receive projection from SC, why there is a lag about ~60ms between the responses in the two sites.I have another concern about one technical detail. They measured DA concentration with amperometry, which is the current pass through the electrode to oxidize dopamine. They addressed the specificity of the signal by ruling out contribution from 5-HT or NE, but I am not sure whether the transient response they saw was actually a reflection of evoked electric potential at the recording site, especially they averaged over the trials to extract the DA transient.

Neural Representation of Task Difficulty and Decision Making during Perceptual Categorization: A Timing Diagram

2006-08-31T09:16:00.000-04:00

Roger Ratcliff is a well-known proponent of the diffusion model of decision-making, who I think is one of the few people with good interesting ideas on the topic.

Here, he consults with Pliliastides and Sajda to help follow up a recent discovery they made about ERP and decision-making.

They had earlier (Cerebral Cortex) shown that a specific ERP component very closely predict behavioral performance in a classification task. IE, they made neurometric functions and found high correlation for this component. It was at >300 ms.

The timing, but not the intensity, of this component depended on task difficulty. This suggests that in more difficult tasks, the brain thinks more and reaches a decision more slowly. How does the brain know it needs to think more?

Here, they authors go back to this data and identify a component that arises just before the 300 ms component that has an intensity that is correlated with task difficulty. They call it the D220. The strength of the D220 increases with increasing difficulty.

To control for stimulus effects, they do a nice little control: they vary the background color, and on some trials, subjects have to classify the background color. So now they can manipulate how scrambled the image is, which controls the difficulty of the identification task, and they can change the color of the image. So they can manipulate the stimulus independant of difficulty. Sure enough, D220 varies with difficulty, not image properties.

Unfortunately, the authors do not address the issue of localization at all. Many people would suggest that Anterior Cingulate monitors for conflict (which is roughly the same as difficulty) and induces executive proceses to occur when conflict is high. This idea has recently been challenged and its status is currently ambiguous. This study doesn't really help us either way with the question of ACC function.

The main contribution of this study is to divide the decision process into temporally discrete components, which can serve as a foundation for later studies that identify the neural substrates of these processes.

Fast modulation of prefrontal cortex activity by basal forebrain non-cholinergic neuronal ensembles

2006-08-24T23:12:00.000-04:00

Ok, here is a shameless plug of my very own paper.

Some background: Most of us know basal forebrain (BF), also called Nucleus Basalis of Meynert, as the center of cortical-projecting cholinergic (ACh) modulation. But a well concealed secret is that there are more non-ACh (mostly GABAergic) neurons in the BF that project to the cortex. Never heard of it? So did I when I started this project. The question is what do they do?

This paper provides some good data (and some novel analysis techniques) to support the idea that one electrophysiologically-identified class of BF neurons, putatively non-ACh, is capable of fast (~200 msec) cortical modulation. These BF neurons transiently synchronize, leading to activation of prefrontal cortical network and brief increases of gamma oscillation.

The result in itself supports a novel fast cortical modulation mechanism, mediated by a diffuse-projecting subcortical system. This is the opposite of the general idea that neuromodulatory systems, through their metabotropic actions, are slow-acting.

What is this non-ACh BF system good for? I propose that this system could serve as a neural mechanism of attention, to transiently amplify cortical representations of behaviorally-relevant stimuli.

Read more here.

2006-08-21T10:45:00.000-04:00

1. When asked to choose between a smaller reward delivered sooner (SS) or a larger reward delivered later (LL), most animals will choose the smaller/sooner reward. This trait is known as impulsivity, and is irrational.

2. Winstanley et al discovered in 2004 that lesions to OFC decrease impulsivity. Very weird. Lesions usually make you more irrational.

3. Hypothesis: OFC is essential for devaluing delayed rewards. Critical test: record OFC responses in a delay/discounting task, see if responses to delayed rewards are lower.

4. Yes.

5. Additional question: do responses depend on reward size? Yes, but these effects are INDEPENDENT. THIS IS THE CRITICAL RESULT OF THE PAPER.

6. These results indicate that OFC does NOT serve as a "common currency" area of the brain (this contradicts Montague+Berns and Padoa-Scioppa+Assad).

7. Also, OFC is direction-tuned. This is very novel.

8. Collectively, these results indicate that OFC is NOT a common currency area. But instead it is kinda a lot like PCC, ACC, PMD, etc. Just another typical reward/premotor area.

Check it out!

An RNA gene expressed during cortical development evolved rapidly in humans: Pollard et al, Nature, 16 August 2006

2006-08-20T21:59:00.000-04:00

Studies in the past that compared the genome of humans with those of other animals have focused on the comparison of genes in the strict definition of the word: a gene is a sequence of DNA that ultimately encodes a protein. However, 98% of our DNA does not encode proteins. In this current study, they scanned these non coding regions in order to isolate human accelerated regions (HARs), regions that showed an accelerated rate of subsitution since the divergence of humans and chimps. The most amazing finding was a region they termed HAR1, consisting of 118 bp. "Only two bases (out of 118) are changed between chimpanzee and chicken, indicating that the region was present and functional in our ancestor at least 310 million years (Myr) ago." However, comparison of this region against a 24 person diversity panel shows that 18 out of the 118 bases have been substituted in humans since our divergence from chimps, and, furthermore, that these changes were likely to have occured within one million years ago.

Okay, so we have a section of DNA that's changed dramatically since chimps -- what does it do? Although it does not encode a protein, models indicate that it could transcribe a stable RNA structure. In situ hybridizations revealed that the RNA is present in the embyronic human brain starting at 7-9 gestational weeks. It is present in the dorsal telencephalon (the developmental precursor to cerebral cortex) but not other parts of the forebrain. Coexpression with markers for the protein reelin showed that HAR1 is present in the Cajal–Retzius neurons, the scaffold structure that allows migrating coritical neurons to wriggle into place.

Analysis of adult brains showed that the RNA is present mostly in the cerebellum but "is also prominent in forebrain structures, including the cortex, hippocampus, thalamus and hypothalamus." Oddly, it also showed up in the ovaries and testes, but not in any of the other tissue samples tested (skeletal muscle, thymus, liver, etc.).

Frames, Biases, and Rational Decision-Making in the Human Brain -- De Martino et al. 313 (5787): 684 -- Science

2006-08-15T21:56:00.000-04:00

Human do not always make rational decisions, an example is the frame effect these authors study. Subjects are initially given $50 and subsequently asked to choose between a sure option of keeping $20 or gambling for $50 with a 40% winning probability. The expected values for the two choices are the same, as most subjects clearly recognized. Nevertheless, subjects are more likely to make the sure option when prompted “Keep $20” and are more likely to gamble when prompted “Lose $30”. These biases are interpreted as emotion modulating the decision making process, and let me call them “emotionally intuitive” choices. Thus for a given subject, the percentage of emotionally intuitive choices is slightly more than half. This percentage differs across subjects, some subjects are more ‘rational’ and the percentage is closer to 50%.

What are the neural bases for this behavior? To identify brain regions responsible for this behavioral bias, the authors compared brain activity between emotionally intuitive vs. counter-intuitive choices and found that amygdala is more active for emotionally intuitive choices while ACC is more active for emotionally counter-intuitive choices. To further identify brain regions responsible for the individual differences (how rational each subject is), they found orbital and medial prefrontal cortex (OMPFC) is more active in more rational subjects. Thus the results would suggest that OMPFC activity does not differentiate between emotionally intuitive and counter-intuitive choices, but the tonic OMPFC activity throughout the task is correlated with how rational the subject is. Conversely, rational behavior does not correlate with a higher or lower Amygdala activity. Together, these results suggest that OMPFC is regulating emotional information carried by the Amygdala.

As an electrophysiologist, what I am most interested is what are the underlying neural activities that give rise to the BOLD signals in this network, and how does the cross-region interaction/regulation take place? On the other hand, this reminds me of the limitations of chronic in-vivo electrophysiology: although we can directly access neuronal activity at the single neuron level, we lack the proper description of neuronal activity at the population level that can be easily and objectively compared across subjects.

Midbrain dopamine neurons encode decisions for future action - Nature Neuroscience

2006-08-07T18:12:00.000-04:00

Genela Morris, Alon Nevet, David Arkadir, Eilon Vaadia & Hagai Bergman

This paper is succinctly summarized in the News and Views article by Niv, Daw and Dayan, which is also a lot more readable than the paper itself. The experimental design is quite clever, combining single-target reference trials and double-target decision trials to study how dopamine (DA) neurons respond under the two conditions. The main question is, when faced with two alternative choices, what does DA neurons signal? The average value of the two stimuli? The value of the better choice? Or the value of the chosen action?

The answer is that DA neurons reflect the value of the chosen action, in as little as 120msec time after stimulus onset. This result indicates that decision signals, on average, are relayed to the DA neurons very early on. Therefore, DA modulation does not act to bias decision (because the decision has been made) but used for something else.

The beauty of studying DA neurons, to a large extent, is the computationally precise model. These models adequately describe behaviors and neuronal responses and provide quantitative predictions that can be directly tested. The mathematical formalization, and the insights required to generate these models, are the essential element needed to further understand other neuromodulatory systems.

Activity of striatal neurons reflects dynamic encoding and recoding of procedural memories : Nature

2006-08-07T14:00:00.000-04:00

This is a paper from A.M. Graybiel group. They recorded striatalsensorimotor neurons in rats during acquisition (learning),over-training, extinction and reacquisition of a T-maze task. Theybasic hypothesis is that the basal ganglia promotes variability inbehavior during trial-and-error learning (exploration) and evaluatesbehavioral changes leading to "explotation", the optimal behavior,in which correct choices are made consistently. They argue that theactivity in basal ganglia should correlate with this behavioralphenomenon. They look at different global properties of activity ofstriatal neurons as a function of the task stage (acquisition,over-training, etc). Such global properties are: population averagefiring rate, entropy, and spike progression index (the correlationbetween the firing rate vector of any stage with the firing rate vectorof the over-training stage), among others. These propertiesconsistently change depending on the task stage, exhibiting the highestvalues during over-training, decreasing during extinction and fastregaining high values during reacquisition. The most striking feature,in my opinion, is that during over-training, striatal neurons increasetheir firing rate at the very start and at the end of the trials, likemarking the time limits during which the animal is actively performing(and probably concentrated) the task. What I missed from this paper isthe percentages of neurons behaving like this; neither the total numberof recorded neurons is given. They claim it is because they probablyrecorded repeated neurons in different days, thus, give a total numberwould be inaccurately. Knowing the proportion of task-responsiveneurons during each session would be good enough. Tetrodes are used forrecordings, but most important analyses showed in the paper use averagesite or global activity.

Anterior cingulate cortex responds differentially to expectancy violation and social rejection - Nature Neuroscience

2006-08-06T10:08:00.000-04:00

You look at a photo of some random person, and find out whether they like your photo.

If they say they like you, you get a big BOLD response in vACC.

If not, no change.

Social acceptance is mediated by vACC! Phrenology!

Actually, this very short paper is designed to clear up confusion arising from an earlier paper that found social acceptance/rejection in dACC. Traditionally, dACC is thought of as the cold calculating logic area, and it is thought to handle general conflict. (It's connected with PFC). Conversely, vACC is though of as the emotional and social part of ACC. (It's connected to OFC).

So this earlier study was confusing to everyone.The present study clears up the problem by showing that expectancy violation is what is coded by dACC in the previous study and that social rejection per se is encoded by vACC.

All I know is, next time I feel rejected, I will know where all the blood in my brian is not going.

Encoding and Decoding Touch Location in the Leech CNS -- Thomson and Kristan 26 (30): 8009 -- Journal of Neuroscience

2006-08-03T08:34:00.000-04:00

This paper proves again that the leech is not as lame a system as most people (like me) tend to think.In this paper, the authors poked leeches on the side and recorded responses of sensory neurons. The leech only has two relevant somatosensory neurons (!), so they authors could record from ALL relevant neurons. They found that response latency and spike count account for all the information in the spike trains. In other words, anything else in the time domain didnt count. They did not test synchrony, although they probably have the data to do so. Interestingly, latency carried more information than number of spikes.I and several other people I know like Bill Kristan's work because it seems relevant to primate work, and this is a good example. There is a long history of debate about the relative importance of latency to first spike (or first few spikes) in coding sensory stimuli. The arguments in favor of latency were made most strongly by Van Rullen and Thorpe '97, but primate physiologists seem to have generally ignored this argument. I wonder why. We all have the right data sitting around, we just have to analyze it.In the second half of the paper, the authors stimulated the two neurons using different trains of inputs, and recorded the bending of the leech. They found that bend correlated more closely with rate than with latency. They did not really figure out why this disagreed with the recording, but in any case, latency and rate both carry information. Either way: (1) leeches have a use other than drawing blood, (2) latency is sometimes more informative than spike count, (3) i should go back to my grad school data and see if there is any kind of story in the latencies.

Read more at www.jneurosci.org/cgi/c...

Be careful about the google posting utility

2006-08-03T08:27:00.000-04:00

Be aware! The google commenting utility is not very stable. It has happened to me a couple of times and I imagine that it will also trouble some of you. The post might disappear after you submit for posting. If it works normally, it usually appears within a minute, and you receive receive an email notification instantly.

Therefore, I recommend you to backup your comments before you submit it, or your effort might go to waste.

Neuron -- Sjöström and Häusser

2006-08-01T13:29:00.000-04:00

A Cooperative Switch Determines the Sign of Synaptic Plasticity in Distal Dendrites of Neocortical Pyramidal Neurons

This is a technically sophisticated paper that combines several different techniques: quadruple simultaneous patch-clamp recordings, dendritic patch-clamp recordings, two-photon imaging, and modeling of reconstructed neurons. I like this paper not only because it adds another twist to the LTP and LTD story, but mostly because of its implications on information flow in the cortex. As the authors suggested in the last section of the discussion, the gradient LTP and LTD they observed in pyramidal neurons imply that distal synaptic inputs will be generally weakened. Only the synchronous distal inputs that are strong enough to drive the post-synaptic neuron would be strengthed. Coupled with the fact that top-down cortico-cortical projections terminate in L1 (distal inputs to L5 pyramidal neurons), it would suggest that top-down modulation is mediated through synchronous firing. Multi-electrode recordings will be required to study this topic.

A welcome message

2006-08-01T10:23:00.000-04:00

As I have invited a few more friends to join this site, the intention is to make this blog a pointer toward recently read interesting paper/news/ideas. Since we are all tracking journal articles as they become available, it would be great to share those info with each other so that we don't miss exciting new discoveries in the field of neuroscience.

Here's how this site works:
1. If you're interested in becoming a contributor to this site, please email me first to set up the account.

2. I recommend download this useful commenting tool from google to use with Firefox browser.
This little tool adds a little icon at the bottom of the browser and allows you to comment on any webpage (like a new journal article), which automatically becomes a new post on this site. The recent posts I made all used this function. It's really easy and useful.

3. As a way to notify all members of newly added posts, I created a google group Neurosci_blog with the emails of all participating members. New posts will send alerts through this mechanism. If you'd prefer to read new posts on the website instead of being bombarded with emails, you can change your notification setting in the google group website.

Finally, if you just happen to stumble upon this site, you're most welcome to read it and share your thoughts and ideas with us. If you are interested in contributing posts or receiving email notifications, please drop me an email.

I hope this effort will be worthwhile for all.

-sc

Cortex Is Driven by Weak but Synchronously Active Thalamocortical Synapses -- Bruno and Sakmann 312 (5780): 1622 -- Science

2006-07-27T23:52:00.000-04:00

The paper seems like a technically challenging achievement. The claim that thalamocortical transmission efficiency is achieved through synchronous thalamic inputs is great. It adds to the importance of synchronous firing of neuronal ensembles in information propogation.

Contour Saliency in Primary Visual Cortex

2006-07-06T02:22:00.000-04:00

This should be a good journal club article.

This and many other paper have shown repeated that single neurons in the cortex performs as reliably as the animal's behavior. This observation would be consistent with the idea that single neuron observations are actually reflecting the collaborative interactions of a much larger network. Observing one of them would reveal most information about the entire network (ref to the recent Bialek paper, correlated firing pattern of retinal neurons).

This result also has interesting implications about the neural mechanism of attention.

Role of Substantia Nigra-Amygdala Connections in Surprise-Induced Enhancement of Attention -- Lee et al. 26 (22): 6077 -- Journal of Neuroscience

2006-06-03T13:50:00.000-04:00

Peter is someone I respect a lot. He is extremely smart and has great insights on the neuronal organizations that subserve the intricate behavioral learning phenomena which he is a real master. In fact, my neuroscience career is partly inspired by him. When I started graduate school at Duke, I took his class on advanced learning theory before he left for Hopkins where he is now. Little did I know then that I will be working on something related to his work now.

Back to the paper, also check out this recent one from him.

Read more at www.jneurosci.org/cgi/c...

Temporal Encoding of Place Sequences by Hippocampal Cell Assemblies

2006-04-25T19:56:00.000-04:00

Dragoi and Buzsáki, Neuron 2006

Another paper from Buzsaki demonstrating the interaction between hippocampal place cells. Place cells are more tightly coupled with each other than predicted by common inputs (place information). How is the result presented here different from the Harris (2003) Nature paper?

An important result of this paper is the demonstration that cell assemblies corresponding to a behavioral sequence, running tracks in this case, are temporally compressed and represented in a single theta cycle. The temporal sequence of different place cell assemblies within each theta cycle (phase) represents the larger time-scale activation sequence of these neurons.

This paper, as well as Schneidman et al paper (pairwise correlations), are starting to elucidating that, in the in vivo environment, neurons operate as cell assemblies cooperatively, but not as independent individual neurons.

Noradrenergic Activation Amplifies Bottom-Up and Top-Down Signal-to-Noise Ratios in Sensory Thalamus -- Hirata et al. 26 (16): 4426 -- JN

2006-04-25T19:41:00.000-04:00

Contrasts the effects of NE and ACh modulations on thalamocortical transmissions. Consistent with many previous results, show that NE reduces baseline spontaneous firing rates and results in sharpening of signal-to-noise ratio.

Compare to results from Devilbiss and waterhouse (2004).

What is the implication of this effect at the population level and on a single trial?

Read more at www.jneurosci.org/cgi/c...

Recurrent Connection Patterns of Corticostriatal Pyramidal Cells in Frontal Cortex -- Morishima and Kawaguchi 26 (16): 4394 -- Journal of Neuroscience

2006-04-25T17:40:00.000-04:00

This is a nice study combining in-vivo regrograde labelling with in-vitro characterizations of the connectivity of the two different projection cell types. This is a good combination of techniques I would like to emply in the future.

Read more at www.jneurosci.org/cgi/c...

A Spin Glass Model of Path Integration in Rat Medial Entorhinal Cortex -- Fuhs and Touretzky 26 (16): 4266 -- Journal of Neuroscience

2006-04-25T17:35:00.000-04:00

This is a nice modeling paper trying to explain the beautiful hexagonal spatial map recently discovered in the entorhinal cortex by the Moser group. Could be a good journal club paper.

Read more at www.jneurosci.org/cgi/c...

Reverse replay of behavioural sequences in hippocampal place cells during the awake state : Nature

2006-04-25T17:20:00.000-04:00

Foster and Wilson nicely demonstrated that hippocampal place cell populations display reverse replay of behavioral sequences right after rats finished running a linear track and stopped to rest. Instread of foreward replay that has been previously documented during both stages of sleep, especially during sharp wave events during SWS, these reverse replays occured during WK, also concurrent with sharp waves.

The mechanism underlying the reverse replays, as suggested by the authors, could be the traces of neuronal excitability leftover from the track-running. Thus during sharp wave events when the inhibition tone is decreased, the last neuron active during the sequence fires first because of its highest excitability. The implications of the reverse replay could be to facilitate backward associations of learned sequences. The authors suggested that the reverse replay could be used to allocate synaptic changes in accordance with the temporal difference model, provided that the reverse replays are temporally coincident with burst firings of DA neurons. This assumption is an emperical question that can be tested.

I think this study nicely illustrated the advantage of studying hippocampal place cells, which offers a behavioral context where the spiking sequence of neuronal populations is meaningful and interpretable.

Weak pairwise correlations imply strongly correlated network states in a neural population

2006-04-24T23:50:00.000-04:00

This is quite an interesting paper, both conceptually and analysis-wise. The main point of the paper is that the information content of the ensemble spiking pattern (in this case retinal ganglion cells) can be mostly described using only up to the second-order interaction between neurons, i.e. pair-wise correlation. Unlike independent Poisson models, the existence of pair-wise correlations indicates the relative abundance of synchronous population activity among neurons. Quoting the last sentence of the paper:

In this view, the network is much more than the sum of its parts, but a nearly complete model can be derived from all its pairs.

The statistics of the ensemble spiking pattern can be parametrized with N(N+1)/2 parameters, which can be implemented and decoded by the neuronal network with local synaptic learning rules. As the number of neurons grows, the spiking pattern can be retrieved given partial information of ~200 neurons (Figure below).

The Implications of this result are several fold, as stated in the paper:

Finally, whereas the maximum entropy model can be constructed solely from the observed correlations among neurons, the conditionally independent model requires explicit access to repeated presentations of the visual stimulus. Thus, the central nervous system could learn the maximum entropy model from the data provided by the retina alone, but the conditionally independent model is not biologically realistic in this sense.

The issues that were only partially addressed and remain to be fully developed are (1) how does the ensemble spiking pattern interact with external stimuli? (2) how does the pattern evolves through time? The latter question could be resolved using techniques like locally-linearly embedding as in Stopfer et al.

Another theta oscillation paper

2006-04-16T13:55:00.000-04:00

1. Sri's paper from Lisman lab. Using cortical surface grid electrode to map out LFP responses during working memory tasks in human. The main point of the paper is that cortical theta oscillations are locally generated. Although many cortical areas are recruited during the task, they are not coherent nor form distributed networks.

2. Sri discussed about the discrepancy between EEG studies and iEEG results he showed. It could be that EEG studies retult from volume conduction. But that same argument may not be used account for gamma-band coherence and there are many of those studies (Fell, Fries, Engel, Varela) demonstrating that effect.

3. Compared to the Tsujimoto study where bipolar recording was used to assess local theta generation, this study used unipolar recording. Based on this result, it can be argued that (at least in human), volume conduction is not a big issue. Unlike the current study where long-distance coherence was not common, the Tsujimoto study showed theta coherence between the midline PFCx and ACC.

4. These short-lasting theta modulations (along with phase resetting) could be generated by a common event-like modulation (BFTN?)

Theta Oscillations in Human Cortex During a Working-Memory Task: Evidence for Local Generators -- Raghavachari et al. 95 (3): 1630 -- Journal of Neurophysiology

Transient theta oscillation in primate PFCx and ACC

2006-04-16T12:51:00.000-04:00

1. Bipolar recording of cortical LFPs mapped out the source location of midline theta oscillation in the medial PFCx and ACC.

2. Self initiated movement and reward delivery are associated with transient increase in theta oscillation power, lasting less than 1 sec.

3. These external events are also associated with transient theta-phase resetting! Modulated by BFTNs?

Direct Recording of Theta Oscillations in Primate Prefrontal and Anterior Cingulate Cortices -- Tsujimoto et al. 95 (5): 2987 -- Journal of Neurophysiology

Dopamine interaction with MS/vDB neurons

2006-04-16T12:13:00.000-04:00

1. An urethane study in rat MS/vDB, showing that D1/5 can modulate the theta rhythmic firing properties of MS/vDB neurons, in both excitatory and inhibitory directions.

2. Provides some good references regarding the interaction of DA and the MS-Hipp system.

3. Does not address the neurochemical nature of recorded MS/vDB neurons. It remains possible that GABA neurons are involved.

4. One interesting point is that the DA-facilitation of hippocampal ACh release is through their direct actions in the dorsal hippocampus (because D1/5 antagonists locally administered in the dorsal hippocampus block this effect). The question is then: what are the functions of DA terminals in the MS complex and its apparent effects in entraining MS neurons to fire in theta rhythm?

Dopamine D1/5 Receptor Modulation of Firing Rate and Bidirectional Theta Burst Firing in Medial Septal/Vertical Limb of Diagonal Band Neurons In Vivo -- Fitch et al. 95 (5): 2808 -- Journal of Neurophysiology

Three paper from Neuron

2005-04-07T11:38:00.000-04:00

Corelease of Dopamine and Serotonin from Striatal Dopamine Terminals
Fu-Ming Zhou, Yong Liang, Ramiro Salas, Lifen Zhang, Mariella De Biasi, and John A. Dani

[Summary] [Full Text] [PDF] [Supplemental Data]

"Neurotransmitter transporters have long been known to recognize related compounds as substrates, resulting in the accumulation and release of so-called “false transmitters.” In this issue of Neuron, Zhou et al. show that when serotonin levels are elevated by inhibition of either serotonin reuptake or of monoamine oxidase, dopamine neurons accumulate serotonin. The results suggest that release of serotonin by dopamine neurons may contribute to the effects of multiple major classes of antidepressants."

Unitary IPSPs Drive Precise Thalamic Spiking in a Circuit Required for Learning
Abigail L. Person and David J. Perkel

[Summary] [Full Text] [PDF]

"When is an inhibitory synapse not inhibitory? In this issue of Neuron, Person and Perkel demonstrate that thalamic neurons can translate extrinsic GABAergic input from the basal ganglia into highly precise patterns of sustained spiking in a circuit that is essential for vocal learning in songbirds. Postinhibitory rebound serves as a mechanism that preserves precise spike timing information, enabling reliable propagation of activity throughout this pathway. The results have broad implications for basic mechanisms of functional processing in both thalamus and basal ganglia and serve to increase our understanding of how acoustic units of vocal sounds are transformed into motor gestures during the sensitive period for song learning."

Prefrontal Phase Locking to Hippocampal Theta Oscillations
Athanassios G. Siapas, Evgueniy V. Lubenov, and Matthew A. Wilson

[Summary] [Full Text] [PDF]

This one finally gets published, after hearing about it more than two years ago.

Cell assemblies

2005-03-27T17:29:00.000-05:00

Hearing Buzsaki's recent talk at UNC, I read again their cell assemblies paper. By looking at the ensemble firing pattern of simultaneously recorded hippocampal neurons, Harris et al showed that the firing modulations of individual neuron can be better predicted using the spike times of other simultaneously recorded neurons. In other words, in addition to common sensory input modulations (like place fields), hippocampal cells form assemblies that tend to fire synchronously, i.e. they would work together as opposed to firing independently. Interestingly, the time constant for best peer prediction is 10-30ms, the time scale of a gamma oscillation cycle.

This paper provides a good example of asking how the network operates, as opposed to asking how the network subserves its biological function. This network-centered perspective might prove to be a useful alternative to understand the function of the neural network.

Reinforcement Learning

2005-03-20T23:59:00.000-05:00

As I was reading this interesting review about basal ganglia, it made several references to the book Reinforcement Learning by Richard Sutton and Andrew Barto. The book is made available online.
The discussion in this review paper points to an interesting parallel between statistical learning algorithms and how the nervous system implements learning online. Similar ideas were pursued by Peter Dayan on the computational roles of DA, 5-HT, NE and ACh.

Prog Neurobiol. 2003 Dec;71(6):439-73.
Information processing, dimensionality reduction and reinforcement learning in the basal ganglia
Bar-Gad I, Morris G, Bergman H

Reinforcement Learning: An introduction
Richard Sutton and Andrew Barto
MIT press, 1998

Spike-timing-dependent synaptic plasticity depends on dendritic location

2005-03-14T10:05:00.000-05:00

More work on spike timing dependent plasticity (STDP) from Moo Ming Poo. This paper claims to show that the properties of STDP (its strength and the width of the temporal window of influence) depends on where you are along the length of the dendrite. They also have modelling results suggesting that such variation eventually leads to postsynaptic selection of different features of input spike trains depending on where you are along the dendrite.

Just a few more factors for the modellers to think about in a long line from the Poo lab...

******************

Froemke, Poo, and Dan
Spike-timing-dependent synaptic plasticity depends on dendritic location

In the neocortex, each neuron receives thousands of synaptic inputs distributed across an extensive dendritic tree. Although postsynaptic processing of each input is known to depend on its dendritic location, it is unclear whether activity-dependent synaptic modification is also location-dependent. Here we report that both the magnitude and the temporal specificity of spike-timing-dependent synaptic modification vary along the apical dendrite of rat cortical layer 2/3 pyramidal neurons. At the distal dendrite, the magnitude of long-term potentiation is smaller, and the window of pre-/postsynaptic spike interval for long-term depression (LTD) is broader. The spike-timing window for LTD correlates with the window of action potential-induced suppression of NMDA (N-methyl-D-aspartate) receptors; this correlation applies to both their dendritic location-dependence and pharmacological properties. Presynaptic stimulation with partial blockade of NMDA receptors induced LTD and occluded further induction of spike-timing-dependent LTD, suggesting that NMDA receptor suppression underlies LTD induction. Computer simulation studies showed that the dendritic inhomogeneity of spike-timing-dependent synaptic modification leads to differential input selection at distal and proximal dendrites according to the temporal characteristics of presynaptic spike trains. Such location-dependent tuning of inputs, together with the dendritic heterogeneity of postsynaptic processing, could enhance the computational capacity of cortical pyramidal neurons.

Functionalism and its Discontents

2005-03-11T14:48:00.000-05:00

At the end of the epiphenomenalism thread Shih-Chieh asked me what I really think about consciousness, so I'm going to try to share it gradually in bite-sized pieces. I think he wants me to give some answers instead of just saying why certain approaches are wrong. But I think most people really have to see why the usual approaches cannot possibly lead to a fundamental theory before they will consider the apparently implausible alternative that I advocate.

So I'm going to present two main arguments: the first to show that functionalism cannot offer a fundamental theory of consciousness, the second to show that a classical theory of the brain cannot offer a fundamental theory of consciousness.

I) Functionalism is dead. By functionalism (in the context of a theory of consciousness) I mean the doctrine that consciousness results from implementing some particular function in a physical system. The most recent form of functionalism is computationalism, in which consciousness is identified or associated with specific computations (such as planning, decision-making, self-representation, etc.). The notion that the brain is hardware and the mind is just the software, or algorithm, is a version of functionalism. A consequence is that you should be able to create consciousness out of any physical material as long as it has the right functional organization. (Most contemporary functionalists restrict this "multiple realizability" to have certain "brain-like" features, e.g. similar timing.)

Before I argue that it is dead, I need to distinguish methodological funtionalism (which I believe in) from ontological functionalism (which I believe is wrong). Methodological functionalism says that we will learn more about the neurobiology of consciousness by looking for the functions that are associated with consciousness. Of course. But ontological functionalism says that when we figure out exactly what functions are the conscious ones, THAT's as much of a theory of consciousness as we're ever going to get. Basically it's a new set of laws, tacked onto the rest of physical theory, that say, whenver anything carries out computation X, it is conscious and the consciousness represents aspects of that computation.

I guess the way I just described it alreagy gives it a flavor of arbitraryness, but here's why it really can't be a fundamental objective explanation of consciousness. John Searle expresses the point in terms of the distinction between ontological, or objective properties of a system and observer- dependent properties. This distinction appears to me to be crucial for an effective attack on the problem of consciousness. It leads to what Searle called an “obvious point” he "should have seen long ago" (p. 16, The Mystery of Consciousness), namely that a “physical process is computational only relative to some interpretation.” In other words, functions are not objective properties of any physical system, they are just interpretations by particular observers.

Because there is a tendency for some people to think that functions can be objective and observer-independent, let me put it another way. While consciousness seems to be an intrinsic property of our brain, functions are always extrinsic. What do I mean? When you say this object has function X, you mean it has certain relations to other objects or to you and your goals. It might be relations to future consequences of the object (e.g. memory/planning computations enable the avoidance of dangerous animals), or uses that you yourself put the object to (e.g. a hammer helps drive nails and build things), or relations to an evolutionary history (the heart pumps blood, enzymes make simple carbs), or relations to other objects (the strut holds up the bridge). None of those functions are intrinsic, or implied by the object itself.

When I say consciousness is an intrinsic property of the brain, I mean if you induce the an identical physical state in the brain, you will get an identical consious moment, regardless of the context. It doesn't matter if I'm a brain in a vat with current injections to make me perceive this and that, or if I pop out of nowhere with no evolutionary history, or if I'm killed at this instant before my conscious states can fulfil their functions, I'm still conscious right now.

So I conclude: functions are extrinsic, consciousness is intrinsic, so we cannot just equate a function with a conscious state. Don't be afraid of this argument because you think there is no alternative to a computationalist account--please just consider it on its merits!

II) In a Classical Theory Consciousness Must Be Epiphenomenal.
I assume

There is a neural correlate of consciousness (NCC). It exists whenever a conscious moment or perception occurs. For example, consciousness may be conceptualized as a property of a particular body or physical state.
The NCC is distributed in space. Whether the NCC is a neural network or a single protein molecule.
Consciousness is efficacious: it affects the behavior of the physical system. Not only do different conscious states result in different physical states; but further, the consciousness itself, or each quale, has some effect or influence on the ensuing physical state. (Because if C is a property as suggested above, different conscious states result in distinct physical outcomes, but the consciousness per se could be epiphenomenal.) Evidence: conscious contents such as pleasure and pain correspond closely with evolutionarily desirable signals for reproduction and survival: e.g. fire hurts. The implication is that the qualia themselves evolved. If by some form of natural selection, then qualia must be physically efficacious.
Conscious percepts are unified wholes. Or, every quale involves a comparison of multiple “sensations.” For example, a color is a ratio of three cone activations. A shape is a comparison of multiple spatial locations. Abstract concepts also relate different parts of a conceptual system. Therefore a quale represents a unification of some complexity without losing (all) the complexity. Notice especially that such unity is not just implicit in any collection of data, even if there are regularities in the data. For example, a list of pixel color values does not in itself imply an interpretation as a picture of Richard Nixon.

Now suppose further that the NCC is a classical machine, i.e. any local deterministic model of the brain or any other putative NCC at all. But if the physical model is locally deterministic, its entire behavior is entirely and completely determined by local interactions, without any need to refer to any distributed entities (at all; like waves, patterns, resonances, attractors, macro-properties, what have you). Therefore any such patterns or properties, if they exist, are epiphenomenal, meaning they have no physical consequence, violating axiom 3. Therefore I argue (following Stapp and others) that a classical model of the NCC fulfilling the above criteria is impossible.

Now we're getting closer to the approach I favor: Where classical theory is embarrassingly complete, quantum theory has a hole in it the same shape as consciousness. It is not a case of arbitrarily explaining one mysterious thing with another. The poorly understood reduction of the wave function in quantum mechanics does not necessarily involve consciousness, but it leaves a large opening in the physical theory for previously unrecognized influences, especially coordinating influences, to creep in, probably only under special circumstances. This shows that it is at least consistent to conceive of a causally effective conscious state within quantum theory.

I will confine myself to one example of why I think it is a natural fit. In classical mechanics, configurational descriptions of systems in terms of patterns or attractors or waves are optional; in quantum mechanics a holistic description is mandatory. It is generally accepted that the neural correlate of consciousness is distributed in space (this is true whether one considers the correlate to be activity of many neurons or even a single protein). If there is to be a causally efficacious and indeed ontological physical correlate of a distributed consciousness, the only candidate within known physics is an entangled or “non-local” quantum state.

This essential argument has been made clearly by the distinguished physicist Henry Stapp (an article in Psyche or online at http://arxiv.org/abs/quant-ph/9502012) and also by psychologist Ian Marshall. I believe William James essentially articulated it before the discovery of quantum mechanics (in his critique of “associationism” in The Principles of Psychology).

It seems inescapable to me. But (like Searle's Chinese room argument) it does make reference to subjective qualities (i.e. I assume that consciousness is ontologically unified in some sense), so some scientists may feel at liberty to ignore it.

Now, there are immediate practical objections to the quantum possibility, such as the supposed impossibility of sustaining quantum coherence in a hot brain whose functional elements are very large compared to the usual scale associated with quantum effects. Well, quantum effects have been observed at all spatial scales (e.g. terrestrial non-locality experiments out to 50 km; quantum description of neutron stars). With respect to the temperature, I note that the brain is not at thermal equilibrium, so the normal equilibrium considerations about what is likely do not apply. Although there is no complete theory of non-equilibrium statistical mechanics, energy flow through a system is known to be able to organize a system and stabilize it against perturbations (e.g. Morowitz Energy Flow in Biology). In fact, a concrete model of such an effect in biological systems exists, in which a threshold metabolic energy flow through the system condenses the system into a coherent state of dipole oscillations (Frohlich (1968) J. Quantum Chem. 2: 641-649). Measurements of weak coherent light radiating from living tissues, correlated with life processes such as cell division, may support such a model (Popp in Physics Letters A, 293 (1-2) (2002), pp 98-102 or at http://www.lifescientists.de/publication/pub2001-08.htm). This model answers the temperature objection in principle, but was not considered in a recent critique of the quantum brain hypothesis (Tegmark in Phys. Rev. E 61 (2000) 4194-4206 and Science, (Feb. 4, 2000) 287, No. 5454, p791).

I'll stop here rather than confusing the matter by trying to answer every objection in advance. To sum up so far, I've argued that a classical functionalist account of the brain cannot give a fundamental account of consciousness, but that a quantum theory at least allows a consistent description within known physics, of a causally efficatious AND spatially extended physical state.

I have not said where in the brain I think that state might be or how it interacts with the well-known neural machinery of action potentials and synaptic transmission. I haven't talked about solving the hard problem. But I'd like to hear what flaws y'all see in my case so far, and where you're interested in taking the discussion from here...

Cortical plasticity, Nerve Growth Factor, and Cholinergic system

2005-03-08T20:52:00.000-05:00

Basal forebrain cholinergic system is involved in rapid nerve growth factor (NGF)-induced plasticity in the barrel cortex of adult rats

Prakash N, Cohen-Cory S, Penschuck S, Frostig RD.
J Neurophysiol. 2004 Jan;91(1):424-37. Epub 2003 Sep 24.

Ron Frostig's lab has been using intrinsic signal imaging to quantify the extent of cortical activations in response to a single whisker stimulation. This story started when they published the Nature paper in 1996, showing that topically applying BDNF on the barrel cortex "resulted in a rapid and long-lasting decrease in the size of a whisker representation", while NGF application produced the opposite result, leading to expansion of cortical activation.

The current J Neurophysiol paper looked at the mechanisms leading to NGF-induced expansion of cortical whisker representation. They first showed that NGF receptor, TrkA and p75, are both colocalized with ACh fibers. They showed that IgG192-Saporin, a selective immunotoxin that lesions cholinergic neurons (by binding to p75 and get internalized), can block the effect of NGF-induced expansion.

NGF is known to be important in maintaining the basal forebrain cholinergic projection to the cortex. Since it was suggested in other studies that NGF can also induce immediate release of ACh, these results bring up the possibility that cortical network, instead of being the mere recipient of ACh afferent, can recruit the ACh system by producing more NGF and leads to more plasticity. This is also an issue for extracellular recording of the ACh system because NGF-induced ACh release might be mediated through changing release probability or vesicle content size, processes independent of action potential. Thus while the firing rate of basal forebrain ACh neurons remain the same, the cortical output of ACh might be quite different.

This paper also reminded me the contributions of imaging methods compared to extracellular recordings, especially concerning cortical dynamics. Once imaging method can be mounted to awake, behaving animals and acquire useful information (and possible two-photon), imaging methods can provide more insight by seeing the dynamics of the whole cortical network.

Action potentials, while as the major means to deliver information from one neuron to the next, do not provide the information of neurotransmitter release. Synaptic activities, captured by local field potential activities, might be complimentary in that sense.

A final technical note of this paper: It provides several good references for staining ACh system (AChE, ChAT, VAChT).

Great Illusions

2005-03-02T20:24:00.000-05:00

A great visual illusion on the cover of this issue Nature, can you believe that the two moving disks are physically identical? Details of the paper here.

Do you know other great perceptual illusions that you'd like to contribute?

Here's a couple I think of:

Dale Purves lab website: great visual illusions
Change Blindness

Stimulation: strike two

2005-03-02T20:14:00.000-05:00

Two recent paper extended the concept of deep brain stimulation to treat Parkinson's disease and major depression. This goes well with the concept that the disease state is a disorder at the circuit level. Manipulations that brings the circuit back to its normal activity level also improves the behavioral phenotype. And attacking different part of the circuit might have similar outcome: in the case of Parkinson's disease, stimulating either subthalamic nucleus or motor cortex can be effective.

Imaging the spontaneous and evoked dynamics of cortical microcircuits

2005-02-28T16:57:00.000-05:00

Rafael Yuste visited Duke today and talked about his recent paper and some latest developments. In that Science paper, he described spontaneous repetition of activity patterns in mouse visual cortex slice preparation (using calcium imaging) and also in the spontaneous synaptic input patterns in cat visual cortex recording (in vivo). This repetition was shown to have millisecond precision (in each motif) and involved neuronal ensemble with various spatial topography (columnar, laminar, cluster or dispersed). The repetition also occurred at larger temporal scale with songs consisting multiple motifs. While the order of the motifs were preserved during spontaneous repeats, the temporal gaps between repeats were generally shortened, i.e. a fast replay. The motifs drift over time.

In addition, Rafael also compared spontaneous activity patterns in layer IV of cortex to those when a thalamic stimulation was applied. Surprisingly, the thalamus-stimulation triggered pattern matches with the spontaneous activity pattern. Rafael suggested that, since thalamo-cortico projection constituted only 10% input into cortical network (layer IV?), thalamic inputs serve to trigger the intrinsic cortical dynamics that was already present.

Philosophically, instead of viewing cortical cells as responding to certain features of the external world, he viewed those neuronal ensemble motifs as the functional unit of cortical network, which can be active spontaneously or triggered by stimuli. One can think of cortical network as a complicated energy landscape, with each troughs corresponding to a particular activity pattern (motif). He made comparisons to the synfire chain idea by Abeles, the avalanche paper, and the central pattern generator idea in the spinal cord, suggesting that cortical network is working in ways similar to CPG, but more flexible and complicated. But similar to CPG, the whole pattern is a functional unit, and the spontaneous activity of the CPG determines how the system responds to external stimuli.

One potential criticism is that synaptic transmission is thought to be quite stochastic, and might fail to support a precise repeating pattern demonstrated in his work. Yet, with complicated patterns of activity, synaptic transmission can be more deterministic (so can the spike train).

The neuronal motif and dynamics here may provide the adequate variables to describe the dynamics of network, instead of looking at single neurons. The idea of energy landscape is quite intriguing, leading to ideas that neuromodulation may be able to modulate this landscape by, for example, changing the energy barrier between energy well, changing the spatial extent of each motif, selectively amplify some motifs (in different layers).

The bad news is that the spatial extent the motifs were observed was quite local (within 500micron), making extracellular recording less likely to detect similar effects. This awaits to be tested.

Neuroscience 10 years from now

2005-02-27T14:55:00.000-05:00

How will neuroscience research change in 10 years? What are the issues that will be solved in 10 years? And what will remain? What are the technological advances on the horizon? What are the techniques we will be using/learning in 10 years? What are the main challenges facing neuroscience research now and in the future?

When I started graduate school, got interested in neurophysiology, I was reading mostly single electrode studies in monkey, and felt that things progressed slowly over the past twenty years. Many paper were still debating over the benefits of single versus multielectrode recordings and I naively assumed that multielectrode recording techniques will be the technique that I can learn and use for my whole career. Only until recently I started to realize that many techniques are emerging and impacting on neurophysiology. Neuroscience research, most likely, will change dramatically in ten years. It's really important to stop and think about the future. What is to come?

The techniques that can be used to study the intact brain (non-anesthetized, behaving) are limited, including extracellular recording, EEG and functional imaging.

Neurophysiology (extracellular recording) is probably the technique that changed the least over time. Single- and multi-electrode recordings and microstimulations were widely popular and many interesting observations/experiments (eg Doty)were done before the 70', although most of them now buried in the old literature and forgotten as part of the pre-internet library documents. The biggest change in neurophysiology probably came with the rapid upgrade in computation power, making acquisition and analysis of tons of data possible. But it remains to be seen what additional information/insight can be gained by simultaneously recording from 1000 neurons (compared to, say, 50 neurons).

While neurophysiology offers the best spatio-temporal resolution among these techniques and probably comes closest to underlying circuit/cellular/molecular mechanisms, it is quite limited in many ways. While electrical signals are used to transfer information down the axon, electrical signals are transformed into chemical signals at the synapse. The actual effect of a give spike thus depends on the type of neurotransmitter, the amount of neurotransmitter release (presynaptic) and the amount of receptors (postsynaptic) and its clearance kinetics, none of these available to electrophysiologists.

Several recently developed methods can address some of these issues. Electrochemical methods, like amperometry and cyclic voltometry, can detect extrasynaptic neurotransmitter concentration with subsecond resolution. How sensitive they are remains unknown (a good example). Imaging techniques, on the other hand, can assess find structural information, voltage fluctuations, calcium dynamics, hemodynamics and other molecular components when tagged with fluorophore. Efforts are made now to adapt the imaging setup to study brain activity in intact animals, which is a quite exciting development. The limitation here is that only cortical surface can be studied.

Another good combination with neurophysiology is mouse genetics, which offers interesting model systems with altered physiology. Studying these animal models may bring insight to some disease processes, and thus help us better understand the normal physiological processes. This also brings closer the gap between neurophysiology and molecular/circuit functions.

One big question that remains to be addressed is the proper way to study a interconnecting network (with feedback loops). When all nodes are connected, causal relationship is hard to define and may not be applicable at all. What are the relevant dimensions/variables that best describe the network dynamics?

Poll: is consciousness epiphenomenal?

2005-02-25T11:59:00.000-05:00

I have an opinion, but I'm curious about what people think. Does the conscious aspect of conscious brain states affect behavior or the way neurons fire, or could we just as well be zombies? Do qualia have consequences, or are they just side-effects without any functional effects?

Please note that this question is totally separate from the free-will question, which I am not asking.

Neuroscience discussion forum

2005-02-24T21:32:00.000-05:00

Science always benefits from discussion. So why keep science bloging to myself?

If you are reading this blog, you're most welcome to comment on anything and everything.

If you wish to contribute to this blog by posting your thoughts (not just commenting, but posting), please email me and I will add you as a member so you can post.

If you wish to be notified of changes in this blog, please email me as well. I will add you to a google group (neurosci_blog), which serves as the email listserv so that you will be updated by email.

Different time courses of learning-related activity in the prefrontal cortex and striatum

2005-02-24T13:36:00.000-05:00

Nature 433, 873 - 876 (24 February 2005)
ANITHA PASUPATHY AND EARL K. MILLER

Abstract:
To navigate our complex world, our brains have evolved a sophisticated ability to quickly learn arbitrary rules such as 'stop at red'. Studies in monkeys using a laboratory test of this capacity—conditional association learning—have revealed that frontal lobe structures (including the prefrontal cortex) as well as subcortical nuclei of the basal ganglia are involved in such learning. Neural correlates of associative learning have been observed in both brain regions, but whether or not these regions have unique functions is unclear, as they have typically been studied separately using different tasks. Here we show that during associative learning in monkeys, neural activity in these areas changes at different rates: the striatum (an input structure of the basal ganglia) showed rapid, almost bistable, changes compared with a slower trend in the prefrontal cortex that was more in accordance with slow improvements in behavioural performance. Also, pre-saccadic activity began progressively earlier in the striatum but not in the prefrontal cortex as learning took place. These results support the hypothesis that rewarded associations are first identified by the basal ganglia, the output of which 'trains' slower learning mechanisms in the frontal cortex.

Interesting results from Earl Miller, now that he also has a foot in the basal ganlia research. The earlier presence of behavior-related modulation in the BG is quite intriguing to newcomers to this field, like myself. But the idea is already been propsed and cited in this paper

Houk, J. C. & Wise, S. P. Distributed modular architectures linking basal ganglia, cerebellum, and cerebral cortex: their role in planning and controlling action. Cereb. Cortex 5, 95–110 (1995) | PubMed | ChemPort |

Bar-Gad, I., Morris, G. & Bergman, H. Information processing, dimensionality reduction and reinforcement learning in the basal ganglia. Prog. Neurobiol. 71, 439–473 (2003) | Article | PubMed |

The works by Steve Wise and Hagai Bergman are definitely worth checking out.

I like the serial reversal paradigm and the data analysis of this paper. Using explained variance of a simpled ANOVA model, neuronal data of the population can be summaried, even though the data were recorded sequentially (not simultaneously).

The differences in neuronal population was not treated in this paper. The actual interaction between PFC and BG could not be studied, althoug the authors might have the data (not clear from the paper, and not clear if LFPs were collected).

Stylistic issues: Nice three figure paper, with clear main message. All distracting messages were left out in the supplement. Nice and simple setup in the introduction, by describing two opposite views.

How sleep affects the developmental learning of bird song

2005-02-21T16:13:00.000-05:00

Nature 433, 710 - 716 (17 February 2005); doi:10.1038/nature03275
SÉBASTIEN DERÉGNAUCOURT¹, PARTHA P. MITRA², OLGA FEHÉR¹, CAROLYN PYTTE³ & OFER TCHERNICHOVSKI¹

Abstract:
Sleep affects learning and development in humans and other animals, but the role of sleep in developmental learning has never been examined. Here we show the effects of night-sleep on song development in the zebra finch by recording and analysing the entire song ontogeny. During periods of rapid learning we observed a pronounced deterioration in song structure after night-sleep. The song regained structure after intense morning singing. Daily improvement in similarity to the tutored song occurred during the late phase of this morning recovery; little further improvement occurred thereafter. Furthermore, birds that showed stronger post-sleep deterioration during development achieved a better final imitation. The effect diminished with age. Our experiments showed that these oscillations were not a result of sleep inertia or lack of practice, indicating the possible involvement of an active process, perhaps neural song-replay during sleep. We suggest that these oscillations correspond to competing demands of plasticity and consolidation during learning, creating repeated opportunities to reshape previously learned motor skills.

This paper reminds me of the insight paper Wagner et al that came out last year. In that paper, the authors showed that sleep facilitated insight formation through the clever design of a learning task. Similar to this current paper, insight didn't just occur after sleep, but happened after more effort spent in the task. The group that developed insight, suprisingly, was the group that had worse performance with the conventional strategy.

These paper, together with several others, have demonstrated a clear behavioral enhancement after sleep, in tasks such as textual discrimination and motor sequence learning. What remains to be determined are the physiological processes taking place during sleep and the specfic roles of slow-wave sleep and rapid-eye-movement sleep. The observations in these paper provide some constraints on the model.

I was expecting that learning and activity-dependent plasticity would change the network substantially during awake, and was then re-normalized during sleep. In contrast to my expectations, the performance of the network (assessed by song similarity) neither improved nor decayed during the day, suggeting that the network was at least able to maintain a stable retrieval pattern. The performance actually deteriorated right after sleep, at the same time a latent memory trace was created during sleep, but only discovered later when the network was tapped again.

Unraveling the attentional functions of cortical cholinergic inputs

2005-02-21T10:16:00.000-05:00

Unraveling the attentional functions of cortical cholinergic inputs: interactions between signal-driven and cognitive modulation of signal detection.
M Sarter, ME Hasselmo, JP Bruno, and B Givens
Brain Res Brain Res Rev, February 1, 2005; 48(1): 98-111.

A new review from Sarter. The model they proposed is my favorite on how cholinergic system modulates cortical activity. I encountered very similar ideas first in

Golmayo, L., A. Nunez, and L. Zaborszky, Electrophysiological evidence for the existence of a posterior cortical-prefrontal-basal forebrain circuitry in modulating sensory responses in visual and somatosensory rat cortical areas. Neuroscience, 2003. 119(2): p. 597-609.

The other interesting point in this review is the relationship between the DA system and the ACh system, and how the interaction would play a role in the development of schizophrenia.

Chinese Blog

2005-02-21T07:49:00.000-05:00

I set up a new blog here for any blogs in Chinese. I am not sure how often I will use either of these blogs (sure it's a lot of fun). But as it currently stands, I intend to use the English blog as a notepad for random scientific and academic stuff that I come across, as a way of forcing myself to write in English.

Recruiting Weekend

2005-02-20T15:43:00.000-05:00

This weekend our department hosted prospective students for the recruiting weekend. Our recruiting weekend consists of lots of faculty interviews (of course), a faculty seminar, a graduate student seminar and graduate student poster presentations. On the social activity side, we have graduate student-hosted dinner as a relaxed time for our recruits to see and feel life in Durham. On Saturday night, everyone in the department attended a dinner at a prof's house, a traditional event and a climax of the recruiting weekend.

Several things caught my thoughts during this event. As opposed to previous years where we had up to two recruiting weekends and invited up to 30 students, this year the admission committee was more selective and only invited 14 students. On the one hand, the department realized the benefit of being more selective: that is each student receives more attention and thus would have a better impression and a higher chance to recruit them. On the other hand, the decreased number of recruits reflected the tighter budget they have available for the graduate training program. The NIH budget is going so low that everyone in biosciences feels the influence. Several interdisciplinary graduate programs in Duke took a hard hit; either reduce the number of admissions dramatically or even just close down the program.

The NIH budget cut is probably a historical low. To give an idea about the funding situation, the funding cutoff for NIH grant applications has been about 25 percentile for many years during the economic boom. It is projected (or has already happened) that the cutoff is down to about 10 percentile. That means, for many labs that cannot get their grant renewed, they will have to layoff a significant proportion of research personals or forced to close down. Since NIH budget is in the category of discretionary budget, in theory it could go to zero (0). This budget hardship, not surprisingly, will last for the next few years before Bush steps down, and may continue to be felt for many years to come.

A corollary effect of this financial difficulty is reduced funding for international students. Talking to our Dean of Graduate Study this weekend, I finally realized where the budget for international students came from. Most academic institutions here in the states receive NIH training grant to support the graduate program. Theoretically, this funding can only support U.S. citizens or permanent residents. Yet our department was able to support about 3 international students each year. The trick is that our graduate school taxes on the training grant from each program. By submitting part of the money to the graduate school, the graduate school can then determine how they want to spend the money, such as supporting international student. You must see now how the money laundry works: the money is then shipped back from the graduate school to the department to support international student. With the uprising budget cut, this funding source will take a serious cut for sure.

First Post

2005-02-20T11:30:00.000-05:00

Many of my friends are blogging nowadays, it seems like a fun toy to play with. I'll give it a shot.