Neural responses to complex stimuli

Symposium

Neural responses to complex stimuli

Friday, 10 January 2003, 8:45 am-6 pm

2318 Fillmore Street, San Francisco CA 94115

In the last decades powerful computational models have been developed for the responses of neurons in thalamus and in primary sensory areas of the cortex. These models do a reasonable job at predicting neuronal responses to simple stimuli. Can they explain responses to stimuli that are more complex, and perhaps more natural?

This symposium brought together scientists interested in vision and hearing, who have contributed to their field's move beyond simple laboratory stimuli. They presented their results and discussed techniques used to test, judge, and improve models using complex stimuli.

Organizer:

Matteo Carandini, Smith-Kettlewell Eye Research Institute

Participants:

The symposium was open to the public, and drew about 90 participants. Click here to see pictures of the audience.

We apologize to those we could not accomodate due to the small size of our conference room.

Presentations:

Below are titles and abstracts. Some of the speakers have accepted to place the slides of their presentations on our web site (click on talk titles). We are looking into the possibility of making the audio available as well. More information will appear here in the near future (early February 2003).

	Thalamocortical responses to complex stimuli: an overview Matteo Carandini Smith-Kettlewell Eye Research Institute	A brief introduction to the concepts and challenges involved in understanding responses to complex stimuli in vision, hearing and somatosensation. I will provide background for the subsequent talks, covering the main issues of receptive fields and basic models for the three sensory modalities. An emphasis will be put on similarities across these modalities. I will also mention some recent advances from our laboratory, concerning visual responses of thalamic neurons.
	Naturalistic stochastic stimuli for characterizing visual neurons Eero Simoncelli New York University	A well-known hypothesis states that the design and functional behavior of the visual system is shaped by the statistical properties of natural images. I'll review some statistical properties of natural images, describe some of our recent results on characterizing neurons with stochastic stimuli, and then discuss some new methods for generating naturalistic stimuli for probing visual neurons.
	Coding natural and other complex stimuli in the visual thalamus Pamela Reinagel University of California San Diego	It is thought that retinal and thalamic processing produce a neural code in that is efficient for representing natural visual stimuli. I will present experiments that test how the LGN encodes natural and other complex visual stimuli. The most striking feature of the data is the remarkable temporal precision of the neural response, which was not revealed by more traditional visual stimuli. In conclusion I will revisit and challenge some old ideas about efficient coding of natural stimuli.
	Thalamo-cortical transformation of auditory receptive fields: Anatomy and physiology Christoph Schreiner University of California San Francisco	I will discuss recent developments in our understanding of the anatomical projections from the medial geniculate body to primary auditory cortical fields and potential consequences for the transformation of spectral-temporal receptive fields between these stations.
	Spectro-temporal response fields in primary auditory cortex Shihab Shamma University of Maryland	To understand the representation of broadband, dynamic sounds in Primary Auditory Cortex (A1), we characterize its responses by the Spectro-Temporal Response Field (STRF). The STRF describes and predicts the linear response of neurons to sounds rich with spectro- temporal envelopes. It is calculated from responses to broadband sounds with rippled spectral envelopes. We shall summarize how the STRF is measured, interpret its properties, and discuss their implications to the connectivity of the cell within the cortex and to the thalamus.
	How far can we get with a linear model of auditory cortex? Not very... Tony Zador Cold Spring Harbor Laboratory	We have used in vivo whole cell recording to measure the subthreshold responses elicited by complex stimuli (animal vocalizations and music). Using regularization techniques, we estimated the linear component--the spectro-temporal receptive field--of the transformation from sound to membrane potential. Although responses are very reliable, this linear component accounts for only a small fraction of the response variance. I will speculate about a simple nonlinear model incorporating a time-varying adaptation term that may capture more of the input-output structure, and will close with some mumblings about a more general notion of "stimulus complexity".
	The "curse of dimensionality" in the analysis of neural responses to complex stimuli Dario Ringach University of California Los Angeles	The analysis of neural responses to complex stimuli is difficult due to the high-dimensionality of the input space. The basic problem is that of estimating a probability distribution in a high-dimensional space given a finite experimental time. I will review the various methods that have been proposed to analyze the responses of neurons to complex stimuli in the framework of probability density estimation, discuss their relative advantages and disadvantages, and speculate on what techniques and strategies may succeed in the future.
	Analysis of visual coding with complex stimuli Yang Dan University of California Berkeley	A major challenge in studying sensory processing is to understand the meaning of the neural messages encoded in the spiking activity of neurons. In the visual cortex, the majority of neurons have nonlinear response properties, making it difficult to characterize their stimulus-response relationships. I will discuss two nonlinear methods to analyze the input-response relationship of these cortical neurons: training of artificial neural networks with the back-propagation algorithm and the second-order Wiener Kernel analysis. Both methods can capture much of the input-response transformation in the classical receptive fields of cortical complex cells.
	Probing higher vision with complex stimuli Jack Gallant University of California Berkeley	Experiments using simple stimuli and highly controlled viewing conditions have produced good first-order models of early vision, but these classical approaches have been less successful in higher visual areas. We are investigating extrastriate visual processing under more complex, naturalistic conditions. We are also developing nonlinear system identification methods to obtain quantitative models of processing in these areas. This approach may provide new insights about visual function in areas that are nonlinear or are influenced by extraretinal factors such as attention.

S-K home Carandini home