CNI Seminar: Marc Howard

Marc Howard
Theoretical Cognitive Neuroscience Lab
Center for Memory and Brain
Boston University

Scale-invariant representation of time and space in the mind and brain

The world contains useful temporal relationships that are expressed at a broad range of cognitive time scales. In order to be prepared to learn relationships at a range of possible scales that cannot be known a priori, the brain should contain temporal representations of a particular form. At each moment, information about both the time and identity of past events should be maintained on what we can think of as a timeline. Critically, theoretical concerns suggest that the time dimension should be logarithmically compressed, consistent with the Weber-Fechner law from psychophysics. We review evidence from cognitive psychology and systems-level neuroscience suggesting that many forms of behavior make use of such a representation and that the firing rate of neurons in many brain regions show firing consistent with this form of representation. We present a formal model that can construct a logarithmically compressed temporal representation and that can also be generalized to represent other one-dimensional variables such as spatial position, numerosity, and net evidence in decision-making tasks. Time-permitting we will also discuss methods by which this formalism can be extended to construct a logarithmically-compressed timeline of predicted future events.


A pizza lunch will be served.