Abstract.
We have previously characterized the effects of withdrawing attention
on detection and discrimination of static visual stimuli (Lee et al.
Nat Neuro 1998). Here we report attentional modulation of motion perception
in psychophysics experiment. A novel motion stimulus comprising spatio-temporally
contrast-modulated Gabor wavelets was used to distinguish attentional
effects on mechanisms sensitive to component motion from those sensitive
to pattern motion (Schrater et al Nat Neuro 2000). In the second experiment,
we confirmed our component stimulus only activates only early visual
cortex by functioal magnetic resonance imaging (fMRI) measurement, supporting
our argument in psychophysics and consistent with our previous result
(Rees et al. Nat Neuro 2000).
Psychophysics
Methods. Multiple moving wavelets were presented as a composite
patch (1° diameter) presented approximately 4° from fixation. Each wavelet
had a random orientation and phase, and moved orthogonally to its orientation.
A local wavelet mask of different orientation and contrast was superimposed
on each target wavelet. This composite stimulus is expected to selectively
activate component motion mechanisms located in primary visual cortex.
A concurrent letter discrimination task presented at fixation was used
to withdraw attention from the peripheral motion stimulus. Observers
performed a 2AFC target detection task on the peripheral stimulus, while
contrast and local direction of the mask wavelets was systematically
manipulated.
Results.
Novel moving wavelet stimuli was successfully used to characterize
the effects of withdrawing attention on component-motion sensitive mechanisms.
1) Contrast increment thresholds for motion detection showed a ÔdipperÕ
function 2) At low contrast masking, we found direction selectivity
in facilitation effect, which was independent of attention. This sugggests
local component mechanism in early visual cortex is direction selective
and independent of attention. 3) At hight contrast, we found larger
threshold elevation when attention was withdrawn, but not direction
selective. This suggests local direction selective mechanism is dominated
by global mechanism which is not direction selective and dependent on
attention. Global mechanism may be either a) interaction between local
mechanisms or b) mechanism, which integrates inputs from local mechanism,
which locates in later visual cortical area, such as component cells
in MT/V5 which have larger receptive fields with similar property in
cells in V1.
FMRI Imaging
Methods. 4 trained subjects were scanned while they performed
a psychophysical task that is similar to the above in a 2T VISION scanner,
in functional imaging laboratory (FIL) in UCL. Using blood oxygenation
level-dependent (BOLD) fMRI, we characterized the cortical areas sensitive
to component and pattern motion stimulus. Multiple moving wavelets (11°
diameter) were presented 13° to the both sides of fixation point. Subjects
used a pad to indicate the global direction of the moving wavelets of
either side of the stimulus (Direction of attention was directed before
each block and constant during the block). In pattern motion situation
(100% coherence), the wavelets moved coherently and rigidly to only
one direction. Subjects indicated the direction (8° towards left or
right). In component motion (50% coherence), the wavelets contained
50% of wavelets consistent with either up or down, which subjects reported.
Contrast (10% or 80%), coherence (50% or 100%), and location of attention
was manipulated systematically across the blocks (2x2x2 factorial design).
Results. When compared high to low ÔcontrastÕ blocks, we observed
activation difference only in V1, consistent with the past reports.
When compared high to low ÔcoherenceÕ, we observed differential activation
in higher visual areas sensitive to motion (V5/MT and posterior parietal
cortex) but no difference in V1. The results supports our argument in
psychophysics that our component motion stimulates mainly V1 and reflects
the property of early visual motion processing.
