1:00 - 1:20

Lana Yarosh
"Embodied Interaction and Computing"

Embodied interaction is a revolution in the design of computing artifacts, which "emphasizes the primacy of natural practice over abstract cognition in everyday activity." I will discuss the historic aspect of this approach and elaborate with a specific example from my recent research in Computer Science.

1:20 - 1:40

Catherine Zhao
"Learning complex markers from large-scale behavioral data"

We develop computational and experimental methods to gain insights into visual functions and neuropsychiatric disorders. We also build deep learning models that predict human behaviors and identify people with disorders. In this talk, I will discuss our recent innovations on data and models.

1:40 - 2:00

Maria Gini,
with Julio Godoy, Ioannis Karamouzas, and Stephen Guy
"Safe and Efficient Navigation in a crowd"

We propose a Bayesian inference approach where an agent estimates the navigation model and goal of each of its neighbors in a crowd, and uses this to compute a plan that minimizes collisions while driving it to its goal. Simulation experiments performed in many scenarios demonstrate that an agent using this approach computes safer and more time-efficient paths as compared to those generated using a state-of-the-art local navigation framework.

2:00 - 2:20 Break

2:20 - 2:40

Geoff Ghose
"Sparse decoding in perceptual judgments"

In 1972, Barlow introduced a "neuron doctrine" which has had a profound influence on perceptual studies. He argued that sparse encoding on a cellular basis is a basic organizational principle of sensory processing in the cerebral cortex. A necessarily corollary for judgments based on sensory information is sparse decoding, in which the most appropriate neurons for a perceptual task are selectively sampled. I will briefly review electrophysiological recording and stimulation experiments conducting during perceptually based decision making that suggest that such decisions can be based on a highly selective sampling of a small number of neurons.

2:40 - 3:00

Dan Kersten
"The Computational Toolbox for Vision"

Over the past two decades, the computational methods that have been used to model and understand human visual perception have seen a dramatic increase in sophistication and scope. I will provide a brief overview of the field, and then discuss the challenges and opportunities for future research.

3:00 - 3:20

David Redish
"Using computation to identify strategy-related cognition in rats"

Prefrontal cortex in both rats and humans have been identified as a key component for strategy and strategy changes.   Examining prefrontal representations in rats, we find that these representations change after rats are informed that a strategy change is necessary, but before the strategy change is detectable behaviorally.

3:20-3:40 Break

3:40 - 4:00

Bill Beeman
"Hypertext and cognition: A 40-year retrospective"

One goal of American and Northern European higher education is to promote acquisition of a pluralistic cognitive style, which has as an important property--non-lineality. From 1984-7 a research group I directed at Brown University looked at the effects of one of the first advanced hypertext/hypermedia systems,called Intermedia, to develop instructional materials for two university courses intended to promote acquisition of non-lineal thinking. Use of Intermedia was shown to produce significant learning effects enhancing non-lineal thinking. These were somewhat more pronounced for persons involved in developing hypertext materials than for students using the system. In retrospect this was an unusual experimental opportunity to test the effects of the use of hypertext on cognition since the students involved had never before been exposed to this kind of computer application, which is ubiquitous today. In this paper I examine the original experiment and speculate on the lasting effects of the use of hypertext over the past 40 years on cognitive functions.

4:00 - 4:20

Cheryl Olman
"Virtual V1sion: establishing a collaborative computational framework"

This ambitious project seeks to create a common framework in which models of primary visual cortex from diverse labs can be compared directly.  The website that hosts the repository of competing models will also host a broad range of datasets against which model predictions can be tested.  We have two goals:  (1) to level the playing field by making difficult-to-acquire data accessible to everyone, and (2) to move science forward faster by creating a forum in which the community can arrive at consensus on various model elements and build off of each other's work instead of starting from scratch to replicate basic computations in every lab.

4:20 - 4:40

Angus MacDonald, III
"Reward learning versus generalized performance deficits in computational psychiatry: Challenges from the probabilistic reversal learning task in psychosis patients”

A fundamental tenet of computational approaches is that they provide interpretive leverage to understand psychological processes. This assumption may not hold as well when elegant models are used where broad cognitive impairments also exist, such as in patients with psychosis.

4:40 - 5:00 Round Table Discussion