Describe how the Ames Room illusion works
One end of the ‘room’ is larger so people appear to be extremely far away at one end. Binocular view of a room that appears square, but is distorted so it’s bigger along one end. People at each corner of the room: farther away subtend a smaller angle on the retina. We compensate by making the other person much larger since we assume the room to be square.
Determining shape from shading requires a…
model of lighting, as this is actually inherently ambiguous. We resolve this ambiguity by making plausible assumptions about the world (like that light comes from above).
Recognition by Components
One theory of how we put objects together into shapes…
- Recognition based on basic shapes called “geons”
- Objects are defined by the concatenation of geons
- The arrangement of geons is independent of the angle of view, allowing for shape recognition regardless of angle.
A study which seems to undermine RBC theory
Tarr (1995), where subjects were trained on a particular view of a novel object and then tested on recognition of rotated views.
4 problems with Tarr (1995)
- Objects only made up of one geon (cube)
- no obvious parts
- learning was unnatural
- practice was low
Evidence favours viewer/ object centred representations?
Viewer centered
-> object recognition may require experience with many different views (maybe why Tarr was bad)
Agnosia
An inability to recognize some class of stimuli, though the sensory apparatus is intact. It is generally associated with brain injury to a specific region.
Object Agnosia
Inability to recognize common objects.
Agnostic alexia
Inability to recognize text
Prosopagnosia
inability to recognize faces. Often associated with damage to the FFA (fusiform gyrus).
What did the Greeble experiment provide evidence for?
That the FFA is involved in recognizing individuals for domain experts, not simply in recognizing faces.
Describe the problem of selection
-Perceptual systems process large amounts of information automatically and in parallel, and we have to be selective in our processing as our more complex central systems have limited capacity.
3 features of perceptual processes
- Parallel
- Automatic (little control)
- “Unintelligent”
3 features of central processes
- Typically serial
- Often conscious
- Under cognitive control
Filter Theory
- Stimuli are queued after perceptual processing
- Attention switches between input channels (attended and unattended)
- Only attended stimuli are processed further
- We switch our filter between channels depending on what is relevant
Cocktail Party Phenomenon
- Multiple auditory messages available
- Only one is selected and processed
- Analogy of at a noisy bar but only select focus on the auditory message from your friend
Dichotic Listening Task
Shadow an attended message (one ear) and ignore the unattended message
Evidence from dichotic listening consistent with filter theory
Little info from the unattended channel can be recalled, though accuracy on the attended message is high.
Change Detection Task
- Flash picture repeatedly
- Small change introduced between flashes
- Task is to find changing feature
- > have to attend to the correct location.
- > requires focusing of attention as it is not possible when attending to the entire scene.
Change detections shows that coherence of our visual display is ________.
illusory
4 problems for filter theory
- Salient words are detected
- Stimulus coherence matters
- Processing effort matters
- Channels defined by task
When a fluent prose message is switched from the attended to unattended channel and back again (Treisman, 1960), participants follow the _______.
MESSAGE, not the channel!
Detecting Targets Task (Treisman & Geffen, 1967)
- Shadow message in one ear, tap a ruler for “tap”in either ear
- Attended taps always good
- Detection of unattended taps decreases w/ shadowing difficulty
- > hard to explain w/ simple filter theory
