ELSC Seminar Series

Humans have the highly adaptive ability to learn from others’ memories. However, because memories are fallible, for others’ (i.e., ‘witnesses’’) memories to be a valuable source of information, we need to assess their veracity. Surprisingly little is known on how this is done. Thus, I propose a novel theoretical and methodological framework to investigate how we assess others’ memories. The rationale is derived from the notion that autonoetic consciousness—the first-hand re-experience of a past event—is a justification for the veracity of memories. I maintain that from the witness’ side, these justifications are conveyed verbally to share their autonoetic experience (e.g., “I remember Sarah came to the party because I recall she was late and wore a red dress”). Those to whom the memory is communicated (‘receivers’) derive their assessment of its veracity by adopting a third-hand perspective of the autonoetic experience of another—an experience I coin ‘socionoetic consciousness’. Because justifications are communicated verbally, we can use the language of justifications to garner insight on their content and how they are assessed.

I will present the most recent line of work in my lab, looking at how humans assess others’ memories, and whether and how their assessments differ from those of language models. Results of this study show that humans are able to assess the veracity of other’s memories with the same accuracy as language models. Moreover, features generated from humans’ assessments can augment machine-learning language models trained to classify memories.