Theoretical Neuroscience Journal Club

Winter Quarter 2009

January 15, 2009

Mehrdad Jayazeri, Neural correlates of time perception in area LIP during an oculomotor time production task

In the first part of the talk I show data from human and monkey behavior arguing that the internal sense of time in a time production task might be best formulated in a Bayesian framework in which subjective estimate of a sample time interval is derived from (1) the immediate evidence furnished by the sample, and (2) the prior distribution from which the sample is drawn. Second, I will show data from area LIP neurons in one monkey engaged in an oculomotor time production task to examine whether and how the temporal contingencies of an impending time-sensitive saccade might be evaluated during saccade preparation in area LIP.

January 29, 2009

John Tsotsos, York University, Toronto. The different stages of visual recognition need different attentional binding strategies.

Many think that visual attention needs an executive to allocate resources. Although the cortex exhibits substantial plasticity, dynamic allocation of neurons seems outside its capability. Suppose instead that the visual processing architecture is fixed, but can be ‘tuned' dynamically to task requirements: the only remaining resource that can be allocated is time. How can this fixed, yet tunable, structure be used over periods of time longer than one feed-forward pass? With the goal of developing a computational theory and model of vision and attention that has both biological predictive power as well as utility for computer vision, this paper proposes that by using multiple passes of the visual processing hierarchy, both bottom-up and top-down, and using task information to tune the processing prior to each pass, we can explain the different recognition behaviors that human vision exhibits. By examining in detail the basic computational infrastructure provided by the Selective Tuning model and using its functionality, four different binding processes – Convergence Binding and Partial, Full and Iterative Recurrence Binding – are introduced and tied to specific recognition tasks and their time course. The key is a provable method to trace neural activations through multiple representations from higher order levels of the visual processing network down to the early levels.

February 12, 2009

Anne Churchland, Attractor dynamics vs Bayesian inference: A tale of 2 models

Two recent papers have approached the problem of perceptual decision-making from diverging points of view. In the first paper (Furman & Wang, 2008), the authors construct a recurrent cortical circuit model that implements attractor dynamics and attempts to capture as much biophysical detail as possible. In the second paper (Beck et al, 2008), the authors sought to implement principles of Bayesian inference under the assumption that decision-making circuitry is designed to generate optimal choices. Both models represent a clean break from previous models of decision-making in that they use a continuous, rather than a discrete, representation of choices. In spite of their major architectural differences, both models successfully captured some aspects of the behavioral and phsyiological data. The extent to which they represent a major advance over current models will be discussed.

February 26, 2009

COSYNE

March 12, 2009

Guillaume Lajoie: How do spiking networks switch from rhythms to irregularity? Introducing a dynamical systems approach

I will give an introduction, starting from the ground up, to a general set of tools that give insights into the dynamics of networks.

My discussion is motivated by a recent paper by Terman et al.*, which proposes a reduction approach to study mechanisms producing synchrony in networks of inhibitory and excitatory neurons. Such synchronous states arise pathologically, for example, in Parkinson's disease. The key question is how these networks transition from rhythmic, synchronized to irregular, asynchronized states.

Direct numerical simulations yield little insight. However, Terman et al reduce the network to a much simpler map that *can* be qualitatively understood.

Although this is a somewhat technical paper, my presentation is by contrast geared to a mixed audience and is an intuitive discussion of the general task at hand -- as well as an exploration of the benefits of this approach for future investigations in computational neuroscience.

*Transition between irregular and rhythmic firing patterns in excitatory-inhibitory neuronal networks ; J. Best, C. Park, D. Terman and C. Wilson ; J Comp. Neuroscience ; 2007 ; 23:217-235

March 20, 2009 **** Room 328, 3pm ********

Mark Goldman, University of California Davis.

April 9, 2009

Michael Mackey gives the Danz Lecture and may be available for an informal talk

May 6: Special event: Center for Integrative Neuroscience Spring Symposium

Speakers include: Barry Richmond, Wolfram Schultz, Mike Shadlen, Paul Phillips, Sheri Mizumori.

May 7, 2009

Barry Richmond, NIH. Rate and timing codes.

June 4, 2009

David Kastner, Baccus lab, Department of Neurobiology, Stanford University. Distributed Coding Using Distinct Adaptive Modes in the Retina

With their limited dynamic range, neurons have to change their encoding strategy so that they can distinguish their enormous barrage of inputs. Contrast adaptation in the retina serves as a prime example, whereby ganglion cells expand and compress their sensitivity to match the contrast of the input. However, adaptation takes time, and upon a transition from a high to a low contrast the retina is stuck with the decreased sensitivity from high contrast, and fails to respond to the low contrast stimulus. I will present data on a population of ganglion cells that adapt in a distinct way. These cells increase their sensitivity following a transition from high contrast, in effect sensitizing to contrast. In addition, these neurons split the responsibility of encoding the broad distribution of intensities with the standard adapting neurons. The sensitizing cells position their dynamic range to encode the intensities close to the mean—the weak signals—and the standard adapting cells encode the far deviations from the mean—the strong signals—independent of the contrast. I will also present a model of the sensitization based upon evidence suggesting a necessary role for inhibition in sensitization. These results present a new way for a population of neurons to encode a broad and dynamic input by using distinct modes of adaptation to distribute the responsibility.