PostLab
Univerity of Wisconsin-Madison

Memory and Disorders Research Society 2015

ATTENTION AND WORKING MEMORY: A TWO-WAY STREET

 

September 2015
Pembroke College, Cambridge Univeristy

Organiser: Brad Postle

Standard models of attention are mostly concerned with the prioritisation of perceptual analyses according to our current goals. However, top-down biases are not restricted to operate through receptive-field properties, they can also act upon mnemonic respresentations. Furthermore, the mechanisms of selection are, themselves, subject to learning-related modification. This symposium will bring together four lines of work that highlight the complexity of the interplay between attention and memory. First, Kia Nobre will consider how we use long-term memories to project predictions that guide assembly of cohesive perceptions from incoming sensory information. Next, Sharon Thompson-Schill will present evidence that conceptual comprehension is accomplished, in part, by PFC-sourced biasing of activation among the features of conceptual representations in semantic memory. Jarrod Lewis-Peacock will propose a novel account of directed forgetting, wherein an uneven distribution of selection biases neural competition among to-be-remembered items, leading to competition-dependent weakening and, consequently, the forgetting of some. Finally, having previously formalized the operation of selection-from-memory in computational models of “output gating”, David Badre will consider how prior experience with output gating in one context might affect learning and generalization of the dynamics of output gating across other tasks.

Talks:

ORIENTING ATTENTION BASED ON LONG-TERM MEMORY

Kia Nobre, University of Oxford

PUTTING CONCEPTS IN CONTEXT WITH COGNITIVE CONTROL

Sharon Thompson-Schill, University of Pennsylvania

BIASED COMPETITION LEADS TO INTENTIONAL FORGETTING

Jarrod Lewis-Peacock, University of Texas at Austin

SHAPING WORKING MEMORY GATING TO ADAPT TO A DYNAMIC WORLD

David Badre, Brown University 


 

 

 

 


 

 

 

©2017 PostLab.