Protocol

Mouse visual contrast wheel task with temporal regularities

Head-fixed mice report the side of a lateralized visual grating by turning a response wheel while the experiment manipulates trial-to-trial stimulus-side transition probabilities across Neutral, Repeating, and Alternating environments. The protocol preserves sequence regularity as an operational task variable rather than treating it as an inferred history bias.

Task structure

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

Stimulus

evidence_schedule
A grating appears on the left or right screen at one of several contrast levels. Across sessions, stimulus side follows neutral, repeating, or alternating sequence statistics.
evidence_type
static
modalities
visual
notes
—
units
percent contrast
variables
contrast, stimulus side, stimulus-side transition probability, sequence environment

Choice

action_mapping
Wheel turns report the side of the visual stimulus; rare NoGo source rows are mapped to no-response because they are timeout/no-turn outcomes, not a rewarded third option.
alternatives
left, right
choice_type
2afc
notes
—
response_modalities
wheel

Feedback

feedback_type
mixed
notes
Reward contingency is symmetric in the public behavior CSVs used by the MVP adapter.
penalty
Error sound or timeout for incorrect choices.
reward
Water reward for correct left/right choices.

Training

notes
—
stages
wheel shaping, visual discrimination, temporal-regularity environment exposure

Trial timing

  1. quiescence
    protocol-specific

    Mouse holds the wheel still before stimulus onset.

  2. stimulus_and_response
    response-limited

    Visual stimulus is presented while the mouse reports with the wheel.

  3. feedback
    outcome-dependent

Vertical slices

Expected analyses

  • psychometric curve by contrast and sequence environment
  • choice-history kernels by Neutral, Repeating, and Alternating environments
  • reinforcement-learning model with current stimulus and previous-choice memory states
  • dopamine photometry alignment to behavior and model predictions

Interpretive claims

  • adaptive choice-history bias under temporal regularities
    Confidence: medium · ref.fritsche-2024-temporal-regularities

    The atlas keeps environment, previous choices, reward outcomes, and model predictions separable so choice-history bias remains an analysis result, not a hidden task variable.

Open questions

  • Decide whether the IBL comparison CSV in the Figshare deposit should become a separate model-comparison artifact or remain covered by existing IBL slices.
  • Add explicit choice-history kernel extraction after the behavior-first canonical trial table is stable.

Papers

Findings

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

See all findings overlaid on shared axes.