Neuro Tuesday Spring 2001
NeuroTuesday Spring 2001
5-6:30 PM Tuesdays in Warren Weaver Hall Room 1314
January 23:
Judith Hirsch (Rockefeller)
Synaptic structure of orientation selectivity in cat
striate cortex
January 30:
Tony Movshon (NYU)
Control of visual cortical response by contrast and
context
February 6: Aniruddha Das (Rockefeller)
The role of local V1 circuits in complex visual
processing
February 27: Haim Sompolinsky (Hebrew University)
Neural codes and distance measures in probability space
(Note. This is a special, joint seminar with Applied
Mathematics.)
February 28:
Paul Tiesinga (Salk Institute)
Carbachol-induced rhythms in hippocampus
(Note. Special seminar on a Wednesday, but the usual
5 PM start in Warren Weaver 1314)
March 6:
Jonathan Levitt (CUNY)
The anatomical point-spread function of the visual cortex
March 13: NO SEMINAR
Spring Break!
March 20: NO SEMINAR
March 27: NO SEMINAR
April 3:
Anders Dale (Massachusetts General Hospital)
Recent advances in functional brain imaging:
fMRI, MEG, and optical methods
April 17:
Paul Adams (SUNY Stonybrook) Layer 6, synaptic error
and neocortical learning
Abstracts
Paul Adams (SUNY Stonybrook)
Layer 6, synaptic error and neocortical learning
Although different neocortical areas are specialised for different tasks,
all of neocortex follows a conserved bauplan: 6 layers; subcortical input
via thalamus; layer 6 feedback to thalamus; layer 5 feedforward via
thalamus etc. I propose that everywhere layer 6 cells compute a special
quantity which is then used to control the rate at which the feedforward
connections made or received by cortical cells learn. I start by abandoning
the implicit assumption of previous unsupervised or supervised learning
paradigms: that weight updating is anatomically precise. I discuss the
consequence of synaptic learning errors for the simple case of the 1 neuron
principle component analyser. I then show how a novel circuit, using
specialised neurons which measure the temporal synchrony of the spikes of
pairs of synaptically coupled neurons, minimises the effects of synaptic
learning errors. This unorthodox scheme could be implemented by layer 6
neurons. This implementation would require interleaved offline updating of
layer 6 connections using well-defined internally-generated recalibration
signals.
Judith Hirsch (Rockefeller University)
Synaptic structure of orientation selectivity in cat striate cortex
We explore the integrative function of the cortical circuit by
studying its anatomical structure and the physiology of its component
connections. Our approach is to combine the techniques of whole-cell
recording and intracellular labeling in vivo with visual stimulation.
This talk will summarize two projects, time permitting. The first
focuses on understanding the emergence of sensitivity to stimulus
orientation, with an emphasis on the contribution of inhibitory
circuitry. The second asks how the representation of stimulus orientation
changes with iterative stages of cortical processing.
At the first level of cortical integration, in layer 4, simple
cells are able to resolve stimulus orientation and maintain their
selectivity even as stimulus strength increases. The push-pull model of
orientation selectivity holds that the ability to detect stimulus angle
is based on the layout of the simple receptive field, in whose adjacent
subregions stimuli of the reverse contrast evoke responses of the
opposite sign. Thalamic input likely provides the push, or excitation.
The proposed substrate of the pull is a group of inhibitory simple cells.
We have identified such inhibitory neurons; as predicted, their receptive
fields resemble those of excitatory simple cells. In addition, we found
a class of layer 4 complex cells (cells with receptive fields in which
bright and dark responses overlap) that are not tuned for orientation.
These cells could help establish contrast invariance of orientation
tuning and other forms of gain control.
To address the question of whether or not the nature of
orientation sensitivity changes with stage in the cortical circuit, we
measured the relative orientation tuning of excitation and inhibition
across the cortical depth. In layer 4, we and others have found that the
tuning curves for excitation and inhibition share similar peaks and
bandwidths. By contrast, at a later phase of analysis, in layer 5, the
curves for excitation and inhibition often differ, with peaks as far
apart as 90o. Our finding of laminar differences in the relative tuning
of excitation and inhibition shows that the striate cortical circuitry
continually recodes orientation. While such circuit interactions do not
sharpen orientation selectivity per se in cat, they may serve that
function in macaque, where the sharpness and complexity of orientation
tuning increases at advanced stages of cortical integration.
Aniruddha Das (Rockefeller)
The role of local V1 circuits in complex visual processing
The response of a neuron in primary visual cortex (V1) to a simple visual
element embedded within a complex image is generally very different from
the neuron's response to the same element in isolation. This difference, a
specific modulation by surrounding elements in the complex image, is
mediated by short- and long-range connections within V1 as well as by
feedback from other areas. Here we study the role of short-range
connections in this process, and relate it to the layout of local
inhomogeneities in the cortical maps of orientation and space. By using
the measure of cross-correlation strength combined with the optical imaging
of orientation columns we show, first, that the strength of local
connections between cells is a graded function of lateral separation across
cortex, largely radially symmetric and relatively independent of
orientation preferences. We show, next, that in the cortical processing of
complex visual stimuli the contextual influence of flanking visual elements
on the responses of a neuron varies systematically with the position of the
neuron within the orientation map on cortex. The strength of this
contextual influence on a neuron can be predicted from a model of local
connections based on simple overlap with particular features of the
orientation map. This suggests that local intracortical circuitry could
endow neurons with a graded specialization for processing angular visual
features such as corners and T junctions in visual space, and this
specialization could have its own functional map over cortex, linked with
the map of orientation.
Haim Sompolinsky (Hebrew University)
Neural Codes and Distance Measures in Probability Space
How well do neurons represent external events? How big the size of a
neuronal assembly must be in order
to generate an accurate population code? Mutual information and
discrimination accuracy are some of the concepts that can be used to address
these questions. I will show how these quantities are related to the notion
of distance between probabilities. I will introduce Chernoff Distance and
show that in large neuronal populations, this distance rather than the
relative entropy (KL Distance) is the relevant measure for assessing the
quality of the neural code. Examples from neuronal activity in motor cortex
will be presented.
Paul Tiesinga (The Salk Institute)
Carbachol-induced rhythms in hippocampus
Field potential recordings from the rat hippocampus in vivo
contain distinct frequency bands of activity, including delta
(0.5-2 Hz), theta (4-12 Hz), and gamma (30-80 Hz), that
are correlated with the behavioral state of the animal.
The cholinergic agonist carbachol (CCH) also induces oscillations
in the delta, theta, and gamma-frequency ranges in the hippocampal
slice preparation in a concentration-dependent way. I will be
addressing the questions of how the same piece of neural circuitry can
generate oscillations in vastly different frequency ranges and how
neuromodulators can switch the network from one oscillation to
a different oscillation.
I will first describe the experimental results and then present
a biophysical model that reproduces the different synchronized
oscillations as well as the transitions between them as a function
of CCH concentration.
Jonathan Levitt (CUNY)
The anatomical point-spread function of the visual cortex
The responses of neurons in the visual cortex are determined by a variety
of
sources: feedforward inputs from the thalamus or lower cortical areas,
intrinsic
inputs provided by other neurons within the same cortical area, and
feedback inputs from higher cortical areas. However, it has been difficult
to distinguish what these various inputs contribute to the receptive field
properties of cortical neurons. I will describe anatomical studies
specifying the extent of cortical territory providing input to a location
in visual cortex from these sources (what I will call the anatomical point
spread function). I will relate such measures to
physiological measures of the extent of visual space over which neurons'
responses can be driven or modulated, i.e. to the extent of the "classical"
receptive field and of the modulatory surround. I will show that for many
neurons in V1, the extent of monosynaptic feedforward or intrinsic circuits
appears unable to explain the extent of visual sensitivity assessed
physiologically, suggesting that polysynaptic circuits or feedback circuits
must be invoked.
A Recap of NeuroWednesday 2000
Louis Tao,
My CIMS web page,
Applied Mathematics Laboratory &
CNS,
New York, NY 10012. This page last updated Jan 2001.