# Habit Task — Version 2

## Overview

Redesigned habit task following advisor input on the theoretical role of the stimulus (shape) versus context (background) in driving habit formation. Tests whether subjects continue choosing originally-rewarded shapes after a silent reward reversal, with social cues now preserved on social-block probes.

## Changes from Version 1

1. **Background colors removed.** Per advisor reasoning, the shape itself is the stimulus that triggers the habitual response; contextual color was deemed unnecessary because S-R associations should attach to the shape, not the environmental context.
2. **Social cues added to social-block probes:**
   - Training-phase social probes: avatars + reveal of others' choices (full social feedback)
   - Test-phase social probes: avatars visible but no reveal of others' choices
   - Rationale: in v1, social probes had no social information, so subjects had to *infer* what others would choose — that's reflective, not imitative. Adding feedback ensures the imitation process is engaged on probe trials, stamping in S-R associations.
3. **Trial count increased.** 192 total trials (vs. 144 in v1), 12 blocks × 16 trials per block.

## Unchanged from Version 1

- Within-subjects, alternating social and individual blocks
- Silent reward reversal at training-to-test transition (no instructions or notification to subjects)
- Training (blocks 1-6) followed by test (blocks 7-12)
- Probe DV: P(choose originally-high shape)

## Subjects

n=30 (data collection ongoing toward target n=99)

## Result

- Both conditions show test-phase habit above chance (social p=.010, individual p=.052)
- **No social vs. individual difference** (paired t(29) = 0.22, p = .828, d = 0.04)
- Pattern stable across n=15, n=21, n=30 — flat null on the predicted social-habit-enhancement effect
- Training-phase learning weaker in social condition (p=.127) than individual condition (p=.049) — consistent across all sample sizes

## Theoretical Implication

The flat null suggests either (a) the predicted social-habit effect is smaller than the design can detect, or (b) contextual scaffolding (removed when background colors were dropped) was doing more work than the shape-only theory predicted. The training-phase social learning weakness is itself an interesting result — social information may produce noisier learning than individual reward.