# Feature binding and error commission

## TL;DR

The study explores how perceptual features bind with both correct and incorrect responses during tasks, revealing that errors are strongly linked to their associated features.

## Contribution

The research demonstrates that perceptual features bind with erroneous responses regardless of error type or sound timing, challenging previous assumptions.

## Key findings

- Participants more often repeated erroneous responses when the sound was repeated after an error.
- Error type and timing of sound relative to response did not affect the preference for erroneous responses.
- Binding between perceptual features and correct responses appears to depend on contextual factors.

## Abstract

Perceptual and action representations consist of multiple independent features such as color and location of an encountered stimulus, or effector and direction of a performed action. Performing an action further establishes bindings between perceptual and action features, so that reencountering one feature retrieves all bound features. When errors are committed, both erroneous and correct responses are usually strongly represented. In Experiment 1, we investigated the binding between erroneous responses and their effects for different types of errors, with the goal of replicating and generalizing a previous single finding. In Experiment 2, we investigated whether perceptual features bind to correct or erroneous responses depending on whether they appear before or after response execution. These bindings had so far been studied separately. Participants categorized letters via key-press responses, and an irrelevant sound was played after their response (Exp. 1 and 2) or before (Exp. 2 only). Then the same or another sound was played, signaling participants to spontaneously choose a response. After an error in the letter task, participants chose the previous erroneous response more often when the sound was repeated than when it was changed. Surprisingly, neither the error type nor the timing of the sound relative to the response modulated this preference. Thus, the data unanimously support binding and retrieval between perceptual features and erroneous responses. Whether and how binding and retrieval also emerge for the nonexecuted correct response, however, seems to depend on contextual factors and might not be as ubiquitous as has been suggested before.