Protocol

Mouse unforced visual contrast wheel task

Head-fixed mice view one or two Gabor patches on lateral screens and use a wheel to report the side with visual evidence, while trials with no visible stimulus can be rewarded for withholding movement. The protocol distinguishes visual stimulus coding, action initiation, and left/right choice because no-go/no-turn trials are valid behavioral outcomes rather than missing data.

Task structure

Stimulus, choice, feedback, and training as declared in the protocol record.

Stimulus

evidence_schedule
Gabor stimuli are presented on the left screen, right screen, both screens, or neither screen with contrast sampled from a small fixed set.
evidence_type
static
modalities
visual
notes
—
units
percent contrast
variables
left contrast, right contrast, bilateral contrast combination, stimulus side

Choice

action_mapping
Wheel turns report leftward or rightward choices; withholding movement can be rewarded on no-stimulus trials.
alternatives
left, right, no-go
choice_type
free-response
notes
—
response_modalities
wheel

Feedback

feedback_type
mixed
notes
—
penalty
Error feedback for incorrect movement or missed stimulus reports.
reward
Water reward for correct side choices or correct no-go outcomes.

Training

notes
—
stages
wheel handling, contrast discrimination, no-go trials

Trial timing

  1. trial_initiation
    protocol-specific

    Mouse holds the wheel before stimulus onset.

  2. visual_stimulus
    response-dependent

    Lateralized Gabor stimuli are shown.

  3. movement_or_withhold
    response-limited

    Mouse turns the wheel or withholds movement.

  4. feedback
    outcome-dependent

Vertical slices

Expected analyses

  • choice surface by left and right contrast
  • no-go rate by zero-contrast condition
  • reaction time by contrast combination
  • stimulus, action, and choice encoding by trial condition

Interpretive claims

  • perceptual decision-making
    Confidence: high · ref.steinmetz-2019-distributed-coding

    The task is not a strict 2AFC design because no-go/no-turn responses are meaningful and must remain explicit in model comparisons.

Open questions

  • Decide whether no-go trials should be modeled as a third choice or as a censored action state for each analysis.
  • Verify one named extracted session end-to-end and record its subject, trial count, and generated report metrics.

Papers

Findings

Curves attached to this protocol on shared signed-evidence axes.

See all findings overlaid on shared axes.