Will send !

Thanks Toby - looks like it needs to be approved first, must be a new thing from my recollection I thought it went live immediately...

This was my first mentoring journey with my super-talented BA student, Hadas Schiff (not on bsky; congrats Hadas!!!!).

And what a beuatiful day to share this, when the hostages are freed, and finally, there seems to be hope for peace. ☮️🕊️

#computationalpsychiatry

We look forward to feedback!

We plan on building a computational model, and larger-scale test, inspired by work by @thecharleywu.bsky.social , @markkho.bsky.social and others on how we build task representations under uncertainty.

This is small, preliminary but promising evidence requiring replication!

11/12

This helps explain anxiety maintenance: By under-generalizing their behavioral repertoire in threat-related domains, anxious individuals restrict exploration, miss opportunities to disconfirm maladaptive beliefs, and perpetuate avoidance.
10/12

Implication: When an awkward social moment coincidentally co-occurs with your actions (like speaking up when a bell rings), anxious individuals may preclude that behavioral strategy from future task models.
"If I speak up → bad things happen" even when the relationship is spurious.
9/12

This wasn't about reinforcement history.
Only 1 participant in Study 1 hit an obstacle in training. In Study 2, 5 participants hit obstacles (<1% of steps). Removing them preserved all effects.
The bias stems from how threat information distorts task representation, not reinforcement.
8/12

Why does this matter? The bias was differential: Under-generalization hurt performance when facing new instances of threat-related tasks (where that knowledge would be useful) but helped slightly with safe tasks.
Worry predicted worse generalization for threat vs. safe contexts (see Figure 4).
7/12

This is under-generalization, not over-generalization—the opposite of what we see in perceptual fear conditioning!
When making planning errors, high-worry individuals were LESS likely to reuse actions associated with threat contexts, excluding behavioral repertoires from their task models.
6/12

People successfully generalized above chance.
But: Individuals high in trait worry showed systematic under-generalization specifically for the task category that had been paired with threat during training.
Study 1: ρ = 0.43, p = 0.11
Study 2: ρ = 0.53, p = 0.02 (replication with harder task)
5/12

In a planning phase, participants saw NEW vehicles and had to plan 4-step sequences without feedback. Success required correctly inferring the task category of the vehicle.
We tested higher-order generalization ( a car or truck?) and lower-order generalization (which specific car?).
4/12

Participants could easily avoid the fire obstacles during training—and they did! We're not measuring learning deficits or avoidance behavior during training.
Instead, we're asking: Does the mere co-occurrence of threat + actions bias how people later represent and generalize task structure?
3/12

Participants learned to control different vehicles (cars vs trucks) in a grid world. Each vehicle type had unique key mappings defining how it moved—essentially, different "task model" transition functions.
1 type was randomly paired w social threat (fire = more time public speaking😬).
2/12