<html>
<body>
<div align="center"><font size=4><b>The UCSD Department of Cognitive
Science is pleased to announce a talk by<br><br>
</font><font size=5>Angela Yu Ph.D.<br><br>
</font><h4><b>Center for the Study of Brain, Mind, and
Behavior</i></b></h4><b>Princeton University<br><br>
<br>
<font face="arial">Monday, January 28, 2008 at 12pm<br>
Cognitive Science Building, room 003<br><br>
<br>
</font><font size=5>"Sequential Effects: Irrational Superstition or
Sophisticated Computation?"<br><br>
</b></font></div>
Just as visual illusions hint at the principles and mechanisms underlying
natural visual processing, "cognitive illusions" provide
similar insight into the computations underlying decision-making.
In a variety of behavioral tasks, subjects exhibit a sequential effect:
they respond more rapidly and accurately to a stimulus if it reinforces a
local pattern in stimulus history, such as a string of repetitions or
alternations, compared to when it violates such a pattern. This is often
the case even if the local trends arise by chance in the context of a
randomized design, such that stimulus history has no predictive value.
In this work, we use a normative Bayesian framework to show that such
idiosyncrasies may reflect the engagement of mechanisms critical for
changing environments, though not necessarily appropriate for the
randomized 2AFC task. We show that prior belief in non-stationarity can
induce experimentally observed sequential effects in an otherwise
Bayes-optimal algorithm. The Bayesian algorithm is shown to be well
approximated by linear-exponential filtering of past observations,
something also observed in human and monkey behavior. We show that
linear-exponential filtering can be implemented by standard leaky
integrating neuronal dynamics, and that optimal parameters of the process
can be learned without explicit representation of probabilities.
Our work provides a normative account of how decision-making should adapt
to a changing environment, a functional theory for WHY subjects
exponentially discount the past, as well as a principled hypothesis of
how neurons can<br>
implement the necessary computations.<br>
</body>
</html>