Sensory dominance Flashcards
(23 cards)
Gibson (1933) touch straight rod but see it through a prism so it looked curved.
People perceived the rod to be curved (visual dominance).
Rock and Victor (1963) touch vs. vision for shape/size information
When touch and vision provide discrepant shape/size information, vision almost completely dominates perception.
Rock and Harris (1967) impression of size of square in conflict situation vs. touch (haptics) or vision alone
Square judged to be smaller when viewed through a minifying lens. Perceived as a rectangle when viewed through a distorting lens.
Botvinick and Cohen (1998) rubber hand illusion
Participants see a rubber hand being stroked while their own hand (which they can’t see) is also being stroked. When asked to point at whether their hand is, they tend to point towards the rubber hand.
Durgin et al. (2007) rubber hand laser
Around 1/3 of people thought that they felt the heat of a laser when a laser beam was pointed over the rubber hand (could not feel when eyes were closed).
Jackson (1953) visual dominance over audition kettle and whistle or bells and lights
Participants viewed a screen with 7 locations, 30° apart, each of which could emit a sound or a light/steam.
When the deviation between auditory and visual cue was 30° apart, 97% incorrectly responded that the sound was coming from the location visual cue (dropped to 37% for 90°). There was more misattribution for realistic stimuli (kettle and whistle).
McGurk effect
Depending on the speaker’s mouth movement, the same sound could be perceived as either ‘ba’ or ‘da’.
Gallace et al. (2005) receptors for senses
There are more receptors and volume of neocortex for vision (perhaps explaining visual dominance).
Are cows and pigeons visually dominant?
Yes
Posner et al. (1976) on why vision dominates
Visual dominance may reflect the fact that people preferentially attend to vision.
When might audition dominate?
Hearing may be better for temporal tasks (telling us when things happened) while vision is better for spatial tasks (telling us where things are).
Shams et al. (2000) asked how many flashes of light (1 flash in periphery) with auditory beeps
Participants judged that there was 1 flash of light when they heard 1 beep but 2 flashes of light when they heard 2 beeps.
Welch and Warren (1980) modality appropriateness hypothesis
The most accurate/attended sense dominates.
But what does ‘appropriate’ mean? Slightly post-hoc explanation.
Shimojo and Shams (2001) discontinuity hypothesis
The modality in
which stimulation is discontinuous dominates (2 beeps can divide 1 flash, but 1 beep can’t fuse 2 flashes).
Maximum-likelihood estimation definition
To get the most accurate estimate, combine the inputs from each sense by weighting them based on how reliable they are (similar integration for intramodal cues).
More weight is given to the more reliable sense (lower variance). The weight for each sense is proportional to 1/variance (normalised reciprocal variance). Combining the estimate = the estimate of the external stimulus.
3 sensory dominance hypotheses
- Modality appropriateness hypothesis.
- Discontinuity hypothesis.
- MLE
MLE 2 assumptions
- Every sense signal is corrupted by noise (some random error).
- The noise for each modality is independent and gaussian.
Ernst and Banks (2002) 2AFC for which bar presented sequentially was higher (VR so haptic and visual input could be varied independently)
Variable noise was added to the visual signal. As noise increased (0%->200%), visual weight decreased (0.8->0.2), in line with MLE.
By themselves, visual judgements are better than haptic (steeper PF). As noise increases, slope becomes shallower.
There was never >11% conflict between the senses and visual-haptic data always exhibited an influence of both senses. Participants did not note any conflict (some sensory info may be discounted if discrepancy too great?).
Gori et al. (2008) MLE during development
<8yrs, vision dominates orientation judgements and haptic information dominates size judgements.
Statistically optimal integration by 8-10.
Alais and Burr (2004) MLE ventriloquist effect (auditory dominance)
Adding noise to visual stimulus (vision and audition conflict). MLE can explain ventriloquism effect.
Ernst and Bulthoff (2004) on the problem with MLE
MLE tells us how to bind senses, but not which senses to bind. In life, the brain also needs to decided which stimuli to bind (before MLE).
MLE and the binding problem
MLE works well if we know what to combine. We may need Bayesian inference of priors (what is expected) and likelihoods (what is sensed) to work out what to integrate.
Gau and Noppeney (2006) on binding priors with McGurk stimuli
Binding priors are not fixed.
Congruent (matching audition and vision) or incongruent trials. When trial history is mostly congruent, people are more likely to report a fused percept (incongruent = auditory component alone).