How to build a mind without working memory: a computational approach to cognitive architecture
Javier Gomez-Lavin (Purdue University)

February 2, 2024, 12:00pm - 1:30pm
The Center for Philosophy of Science, University of Pittsburgh

1117 Cathedral of Learning
University of Pittsburgh, 4200 Fifth Avenue
Pittsburgh 15260
United States

This event is available both online and in-person


University of Pittsburgh

Topic areas


LTT: Javier Gomez-Lavin

February 2 @ 12:00 pm - 1:30 pm EST

Title  How to build a mind without working memory: a computational approach to cognitive architecture


I want to argue that there’s something deeply wrong with the dominant thread running through models of the mind and its constituent parts and functions, from Aristotle’s emphasis on sensation and the faculty of phantasia, through Ibn Sina’s anatomically anchored ventricular theory, to 19th century cerebral-psychology and its search for the correlates of attention. And that’s that minds need something like working memory to work at all: a capacity capable of hosting disparate elements from across the mental landscape and performing a number of manipulations on those elements in real time. It’s through these operations—which have run the gamut from comparing, evaluating, synthesizing, abstracting, to perhaps most importantly generating new thoughts and memories—that we are said to do much of the interesting cognitive work that humans are thought to do. I’ve argued that there is no one-shot, do-it-all general capacity which in recent years has taken the guise of working memory. Rather, I claim that we perform (most) of these cognitive achievements by dynamically corralling a selection of the many pathways that brains and creatures can take to maintain and manipulate information. Fortunately for us, many of these pathways are clumped and researched under the labels of working memory and other cognitive constructs like executive function. By using simple machine learning processes, we can quickly comb through the extensive working memory literature for candidate pathways. I demonstrate a simplified version of this process that uses LDA topic modeling and highlight some future extensions and possible limitations.

 Can’t make it in-person? This talk will available online with the following Zoom link, found here:

This talk will also be available live streamed on YouTube at

Supporting material

Add supporting material (slides, programs, etc.)




Who is attending?

No one has said they will attend yet.

Will you attend this event?

Let us know so we can notify you of any change of plan.