RT @Ikuperwajs: If you’ll be at #CogSci2025, consider attending our workshop on meta-reasoning! Organized with Lio Wong, @traceymills_, @ka….

8/ If we generate instances of a certain kind by drawing from relevant locations in representational space… why these locations? We suggest they are generally useful in our experiences with the category in question -- more in the paper!.

7/ This holds during decision making as well: people consider options which appear to be drawn from these relevant feature space locations, but select options with different, more context-appropriate features

6/ Category members in these relevant feature space locations “intrusively” come to mind, even when we ask participants to think of those in opposite locations -- suggesting this is a somewhat general mechanism employed regardless of its suitability for the context at hand

5/ If calling category members to mind involves drawing from certain locations in feature space, we’d expect these predictive features to be especially well-encoded in people’s representations of category members. We find that they are!

4/ We find that a category member’s probability of coming to mind can in fact be predicted by its location along certain feature dimensions

3/ We propose that people call options to mind based on the extent to which they have certain category-relevant features. To test this, we empirically identify features people use to represent category members, and locate them w/in the resulting feature spaces

2/ In a variety of familiar categories, we find regularities across participants in which category members come to mind in the absence of a particular goal or decision-making context. So, what makes some category members come to mind over others?

1/ We often find ourselves calling certain kinds of things to mind -- movies to watch, what your guests might want for dinner, names for your pet dachshund (wiggles?)… How do we navigate these relatively unconstrained search spaces?.

Very excited to have this out w/ @phillipsjs! 🧵⬇️.