Ch. 3 - Perception

taking sensory input and interpreting it meaningfully (via inference & learning)

→ main types: visual, auditory, olfactory, haptic (touch) & gustatory (taste)

  • distal stimulus - object/event in real world

  • proximal stimulus - sensory registration (upside down; backwards etc.)

  • percept - meaningful interpretation of proximal stimulus

!! not the same (e.g. size constancy)

  • pattern recognition - categorization of object/event to a class


Gestalt Approach

  • whole > sum of its parts

  • principles of organization:

    • proximity (near) vs similarity (alike)

    • continuation (following continued lines/curves rather than breaks)

    • closure (filling in the gaps)

    • common fate (group of elements move together; e.g. traffic on road)

→ under the law of pragnanz where simple > complex

  • figure-ground organization: distinguishing objects from background

    • is it a baby or trees with their tree branches?

  • subjective/illusory contours: perception not fully determined by stimulus)

    • e.g. phantom triangle made by other shapes


In Research

  • property of emergence: wholes can show new properties as parts combine.

  • configural superiority effect (CSE): when odd elements are detected faster in complex groups than simple groups


Bottom-Up Processing

few sensory input → combined step by step → percept

(built from distal stim. therefore no prior knowledge needed)

→ one way system, no feedback (can’t revise), automatic/reflexive processing

  • template matching

    • sensory input compared to template/pattern

    • near exact pattern match

    • problems: large number of stored templates, variability struggle, only works w/ clean & predictable stim

  • feature analysis

    • recognition when objects → features/parts

    • dog → ears, tail, eyes, paws

    • bierderman’s recognition by components: objects = geons (aka basic 3D shapes)

      • geon + arrangement → recognition

    • supporting evidence: neuroscience, visual search tasks, letter confusions, auditory/speech perception

    • problems: no clear definition on what counts as a feature & there’s too many to recognize an object, misses overall arrangement

  • prototype matching

    • input compared to prototype (most typical representation of a category)

    • doesn’t require exact features; depends on overall similarity (approx. matches)

    • supporting evidence: posner & keele’s dot pattern study (distorted were grouped w/o seeing the prototype), 1999 face recognition study (“prototype” recognized after seeing altered face photos of it)


Top-Down Processing

perception guided by what you already know (context, expectations, past learning)


!! perception is when top-down & bottom-up processing interacts


Marr’s Model of Visual Perception

perception uses special modules which work separately as bottom-up processes

  1. primal sketch - 2D map of edges/brightness/contours; no meaning yet

  2. 2½D sketch - shading, texture, depth

  3. 3D sketch - incorporates top-down knowledge to reconigze objects + give meaning


Perceptual Learning

→ perception improves w/ practice + experience, training attention to the right details

e.g. gibson & gibson’s card task: spot the original card hidden among similarities


Word Superiority Effect

→ being able to identify a letter best in a familiar word than in jumbles or in isolation

(supports that context helps perception)

  • connectionist model (mcclelland & rumelhart)

    • letters & words = many simple units (aka nodes)


Holistic Processing

→ seeing something as a whole, rather than separate features

e.g. laguesse & rossion - holistic face perception

  • composite face illusion: same top halfs of a face look different when the bottom halfs are different

  • other-race effect (ORE): easier to recognize faces of your same race than different due to more perceptual experience


Constructivist Perception

  • perception = ambiguous/incomplete sensory input (proximal stimulus) + learned knowledge regarding stimulus (to fill missing gaps)

e.g. 2D retinal image → 3D mental picture via active interpretation


Direct Perception (Gibson)

  • argument: sensory input provides organized information where mental construction isn’t as required, therefore, perception = direct pick-up of information

e.g. invariance - stable patterns that can be recognizable despite being “different” (such as a melody in diff. key)

  • evidence: point-light displays (johansson) used to show perceptual invariance via dots on key joints + relative motion patterns → perceived as human movement

  • affordance - the possibility for actions that an object offers (dependent on the ability of the object)

    • can be too vague/circular

!! perception linked to action & behavior


Visual Agnosia

→ where people can see but can’t recognize what they’re looking at (able to sense but not perceive well)

*not a memory problem

  • apperceptive agnosia - inability to distinguish visual shapes (able to see outlines but cannot categorize)

  • associative agnosia - inability to recognize what objects actually are (able to categorize/describe)

  • prosopagnosia - face blindness; inability to recognize faces as a whole


Unilateral Neglect (hemineglect)

→ damage to parietal cortex results in ignoring one side of space (usually left side)