implement new stimuli in the stim generator

[mbreiser]

Looming Patterns — Generate looming stimuli with constant velocity or exponential approach profiles
Reverse-phi -- delete current implementation and review papers to get it right, and simple.

and then done, port the implementation to webDisplayTools

[mbreiser]

for reverse-phi @leburnett shared:
the script used to make the reverse phi stimuli can be found here. It uses the function ShiftMatrix which can be found here. You can find the actual patterns here.

When I was making the different versions of the reverse phi patterns I was simply modifying the parameters in this script and then would save the pattern with an appropriate name. The current parameters that are saved are for making a gs_val = 2 version of the reverse phi but the pattern used in the p27 screen protocol is pattern 60 which is a gs_val =3 pattern (Allows for pixel intensities between 0-7. Background value = 3, off = 0 and on = 7). The bars are 16 pixels wide (same as 60 deg gratings) and they move by 8 pixels each frame. The pattern inverts contrast every frame.

This was the protocol I used to test different versions of the reverse phi stimulus. I've attached a screenshot of the turning and centring responses to each stimulus. I've highlighted the condition that was then used in the p27 protocol.

[mbreiser]

Sub-tasks: Looming and Reverse-Phi Patterns

Looming Patterns

• Expanding disc or square from center point
• Two velocity modes:
  • Constant velocity: user specifies step size (degrees per frame)
  • r/v loom: user specifies l/v ratio for biologically-relevant approach timing
• MATLAB implementation exists on a branch (claude/plan-reverse-phy-stimulus-XrTHx), needs testing and validation
• Once confirmed working in MATLAB, migrate to web tools (webDisplayTools pattern editor)

Reverse-Phi Patterns

• Classic reverse-phi motion illusion (perceived motion opposite to physical displacement)
• Current status: make_reverse_phi.m exists but is incorrect — simply inverts alternate frames instead of properly implementing spatial shift + contrast inversion
• Needs reimplementation per literature:
  • Anstis (1970) — original reverse-phi description
  • Chubb & Sperling (1988) — formal analysis
• Key questions: exact timing of spatial shift vs contrast inversion; global vs local inversion
• Action: Either fix implementation or remove from UI until properly specified
• Once confirmed working in MATLAB, migrate to web tools

Migration Path

1. Test and validate looming patterns in MATLAB
2. Fix reverse-phi implementation based on literature review
3. Port both pattern types to webDisplayTools pattern editor (generator.js)