Interleaved stimuli and 3f+4f beat pattern

We can apply our model of cortical motion processing to two psychophysical tasks.

1) Interleaved first- and second-order motion (often seen as evidence for independent first- and second-order pathways)

2) Translating 3f + 4f beat pattern (often seen evidence for luminance and feature based motion processing)


Interleaved stimulus

Made from alternate frames of luminance defined and contrast defined motion stimuli.

Luminance defined stimulus is a translating sine wave grating with additive static binary noise.

Contrast defined (second-order) stimulus is a translating sine wave grating that modulates the contrast of static binary noise.

In each, the envelope translates by a cycle each frame, hence without the integration of motion over consecutive frames the direction of motion is ambiguous.

Psychophysically it is found that subjects can give the correct direction of motion when the frames are all first order or all second order, however when the motion sequence is an interleaved set of first and second order direction discrimination falls to chance levels. This was interpreted as evidence that two separate motion channels existed, and it was argued that any motion system that could see both first and second order motions would incorrectly predict subject performance in the interleaved case 


3f +4f beat patterns

Composed of sum of sinusoids with 3 times and 4 times the fundamental frequency, translating 1/4 cycle every frame (see diagram left)

As the blank inter-frame interval (IFI) increases subjects report that the stimulus is seams to move more in the reverse direction

One explanation is that inserting a blank inter frame interval disrupts luminance based motion processing whilst having comparatively little effect on a feature based motion processing.

 

 


Findings

The model correctly detects the direction of motion in luminance and contrast defined patterns but fails to detect motion when the two patterns are interleaved. These results, shown on the adjacent graph, are in line with the psychophysical data.

With the 3f + 4f beat pattern, the model output contains a mixture of reversed and forwards motion with the latter increasingly prevalent as the blank IFI is increased. This is in line with the psychophysical data.

 

 

 


A PowerPoint presentation from ECVP 1999 summarising this work View Here

Also see: C. Benton, A. Johnston & PW McOwan, Computational modelling of interleaved first- and second-order motion sequences and translating 3f+4f beat patterns. Vision Research. 40 (2000) pp 1135-1142 Download paper in .pdf format

Return to main vision research group page