Some years ago I read about the computational modeling of schizophrenia. (It was probably a summary in a layperson's science publication like Science News.)

As I recall, altering a single variable relating to memory caused a particular type of delusion to arise in the AI patient where, in the recalling of a narrative, characters in the action were substituted with each other, or with the speaker. In the article, this was referred to as an "agent-slotting error".

Web searches for "agent-slotting error" only seem to turn up references to the modeling research, so I would like to know:

  • Is this type of error or delusion something that's actually associated with schizophrenia, or is it just considered a model for understanding aspects of delusions in schizophrenia?
  • If it is something that actually occurs in schizophrenia, what is a term of art for it used generally in psychology rather than specifically in computational models of cognition?

Some clarifying examples from the papers:

From Comprehensive review: Computational modelling of schizophrenia Vincent Valtona:

The hyperlearning model could account for derailments from the original story through a confusion between the characters of different stories (‘agent-slotting errors’) leading to delusion-like ideas (Grasemann et al., 2009; Hoffman et al., 2011).

Grasemann et al. actually calls it “agency shifts”:

Applied to the story and sentence generator networks, hyperlearning robustly produced stable patterns of “agency shifts” where characters migrated between stories and produced meaningful new narratives.

  • $\begingroup$ Could you be referring to source-monitoring error? $\endgroup$
    – Arnon Weinberg
    Sep 17, 2021 at 20:16
  • $\begingroup$ @ArnonWeinberg I don't think it's the same. I'll update the question to include examples from one of the papers. $\endgroup$
    – Theodore
    Sep 17, 2021 at 20:22


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.