Chapter 3

Perception:

Pattern or object recognition

Perception

Sensation vs. perception

What are the mechanisms responsible?

What is the process?

Q: How do we interpret lines and patterns as objects?

Q: How do we program a computer to perceive objects and scenes?

Start simple: How do we recognize these letters as A’s?

Template approach Stimulus is compared to stored pattern

Examples? Bar code, bank check, scantron, etc.

Problems:

There are an infinite number of templates to remember

Have to learn a template first

Any change in stimuli will not be recognized

Receptors in retina -> optic nerve -> occipital lobe (visual cortex)

Specialized receptors in visual cortex

Simple cells: feature detectors e.g. Orientation specific

Complex cells Combination of 2 simple features

Perception due to pattern of neural firing (neural code)

Bottom-up processing

Stimulus

Cell’s

responses

McClelland & Rummelhart (1981)

Interactive Activation Model

Pandemonium (Selfridge, 1959)

Visual perception by neurons

Respond to things that occur most often in environment

e.g orientation: horizontal and vertical lines vs. oblique

Experience-dependent plasticity

Animal reared in certain environment – brain changes to more strongly respond to those cues (Blakemore & Cooper, 1970)

Gauthier et al. (1999): “Greebles” study

Measure FFA (fusiform area)

IV: experience with Greebles

Recognition by components

Biederman’s RBC (recognition by component) theory

36 geons (3D)

Basic building blocks

Emphasis on

intersections

Recognition with missing information possible

Geons:

Identify objects

Principle of

componential recovery

Resistance to visual

“noise”

‘View invariant’

properties

Discriminability

Biederman’s Geons

Intersections are important to recognition

Beyond bottom-up processing

Pattern or object recognition

Bottom-up processing

Information from sensory receptors

Processing driven by stimulus

Data-driven

Top-down processing

Information from knowledge and expectations

Processing driven by higher level knowledge

Conceptually-driven

Problems with pure bottom-up theories:

How does brain pull all the feature information together?

How do theories deal with complex objects?

Tox-Doxn Pxocxssxng

To xllxstxatx, I cxn rxplxce xvexy txirx lextex of x

sextexce xitx an x, anx yox stxll xan xanxge xo rxad xt –

ix wixh sxme xifxicxltx

Context and knowledge fills in the rest!

The redundancy of stimuli provide more features than required

Oliva & Torralba (2007)

Q: Does perception depend on more

than just stimulation of receptors?

Method:

Use same “blob” in multiple contexts

Result:

Perceived as different objects due to top-down processing

Conclusion:

Signal from object

Signal from context

Feedback signal: influence of knowledge

Theory of perception

Bottom-up AND top-down

Bi-directional or connectionist model

Depth perception

Relative size

Size constancy

Odor intensity

Controlled sniff intensity

Perception of language

Speech segmentation

Treisman & Schmidt (1982)

Q: Does knowledge change perception?

Method

Flash display of #s & objects 200 ms

Ask Ss to report #s then objects

IV: Give description of objects (“carrot, lake, tire”) or not

Results

Info significantly improved accuracy

Conclusion

Top-down knowledge changes perception

Able to “bind” features (group information) together more rapidly

Orange & triangle = carrot

1 3

Hollingworth (2005)

Question

How does knowledge of what objects

belong in a scene influence perception?

Semantic regularities (knowledge of function

of objects)

Method

Study scene 20s

IV: w/ or w/o target object

Test: Place target object in scene

By memory or expectation

Result

Accurate position in both conditions

Prediction based on experience

Palmer (1975)

Method Present scene

Ss ID flashed pics (a) or (b) or (c)

IV: type of picture

DV: accuracy

Results Appropriate pictures: 83%

Inappropriate pictures: 50%

Misleading pictures 40%

Conclusion Bottom-up perception interacts with prior knowledge (top-down) to

influence response

Chapter 3

Perception:

Gestalt laws and Heuristics

More perception questions

How does our top-down knowledge system change the way

we perceive the world?

How do bottom-up and top-down information get combined

to provide complete perceptual experience?

How do we integrate all the information (signals)?

Perceptual problem solving

• How do we integrate info into

meaningful whole?

• Helmholtz (late 1800’s): Theory

of unconscious inference and

likelihood principle

• Perception like problem solving

Principles of organization

Gestalt psychology (Koffka, 1935)

Gestalt laws of “perceptual organization”

Pragnanz: Good figure or simplicity

Good continuation

Similarity

Proximity

Common fate

Familiarity

Laws or heuristics (“best guess”)?

Occlusion heuristic

Light-from-above heuristic

Apparent motion: Wertheimer (1912) Method: 2 lights turned on/off in dark room, with right timing,

looks like it is moving (becomes a single light)

Expected result: For larger separations, the stimulus must "move" a

farther distance, which presumably requires a greater length of time.

CogLab

Data from Spring ‘09 (N = 8)

Gestalt law (or heuristics!) examples

Gestalt principles

Group objects together to find the 13 faces

Gestalt law (or heuristics!) examples

Optical illusions and visual phenomenon

Watercolor illusion: bright inside color spreads into enclosed area (b/c center-surround receptive fields?)

Optical illusions and visual phenomenon

http://michaelbach.de/ot/index.html

Motion aftereffect: http://michaelbach.de/ot/mot_adapt/index.html

Lilac chaser: http://michaelbach.de/ot/col_lilacChaser/index.html

Perception problems for computers

Stimulus on receptors is ambiguous

Need top-down knowledge

Segmentation

Visual separation

Speech segmentation

How to deal with visual or verbal noise

Occlusions or obscured

Blurred or degraded

Changes in shadowing (lightness/darkness)

Human perception is different due to bottom-up AND top-down processing

Mitroff & Scholl (2005) http://michaelbach.de/ot/mot_mib/index.html

Method:

Motion-induced blindness (MIB)

Objects superimposed on global motion pattern

Change target objects during MIB

MIB effect

Objects fade from awareness while looking

Mitroff & Scholl (2005) http://michaelbach.de/ot/mot_mib/index.html

Questions:

What happens if change unseen objects?

What is the role of conscious visual awareness in formation

of object representations?

Can object representations not only be formed but also

updated in the face of changing dynamic scenes without

awareness?

Are representations re-formed and updated in response to

unseen visual changes?

Mitroff & Scholl (2005) Method

Ss press a key when experience MIB (dots disappear) and when reappear Choice: simultaneous, one disk, not sure

IV: line changes (50% chance that connecting line changes)

Mitroff & Scholl (2005) Results

Report reappeared simultaneously if dots connected by bar, even if connection occurred during MIB

Object representations can be formed and updated without awareness (conscious perception)

MIB with

Grouping Cues

Result:

More likely to

reenter awareness

together if grouped

together by gestalt

principle!

Current research in perception Presentations: Psychonomic society conference 2011

Metacontrast masking is processed before color-grapheme synesthesia

Action-specific effects are immune to spiders Perception is influenced by ability to act; examined perceptual processing

of spiders (Conclusion: spiders were faster than balls)

Your first organization influences your second: Hysteresis in figure-ground assignment Manipulate how people perceive figures

Pitch height independently modulates perceived time durations

Blindness enhances tactile acuity and haptic 3-D shape discrimination

Relative velocity and relative strength of illusory line motion

Heuristic spatial updating across abrupt perspective changes in dynamic scenes

Color, music and emotion

Auditory perceptual learning through multimodal training

Chapter 3: Perception Research questions

What are the processes responsible for perception?

How do we recognize objects?

Why is perception difficult for computers?

Methods

Identify objects in pictures or read words With or without “noise”

With or without prior information (context)

Indicate what you see with an illusion

Results

Requires combination of bottom-up and top-down processing

Use (gestalt/heuristic) rules of perceptual organization

Experience-dependent plasticity: depends on experiences

Future directions

Chapter 3

Perception & Action

Perception and action

Question: Does interacting with an object change perception?

Examine perception and movement

Example from textbook: how to pick up cup of coffee on table

What are the steps of the process?

What are the mechanisms responsible?

Dissociations Single dissociation

1 function absent while other present

Double dissociation

Opposite pattern in 2 or more Ss

Examples

Broca’s (can’t produce language) vs. Wernicke’s (can’t understand language)

Oliver Sachs’ case study stories

Use: case studies (lesions), imaging, experimental methods

Why is it useful?

Examine if functions are independent

Determine localization of function

Ungerleider & Mishkin (1983)

Method with monkeys:

Object discrimination task

Landmark discrimination problem

IV: lesion site (temporal lobe vs. parietal lobe)

Result:

Object discrimination deficit with temporal lesion

Landmark discrimination deficit with parietal lesion

Milner & Goodale (1995)

D.F.: temporal damage

Visual agnosia: can’t give size, shape or orientation of objects

Tasks (IV: orientation)

Match orientation of card to slot

Put card into slot

Conclusion:

Action intact; perception deficit

James et al. (2003): fMRI study of D.F.

Task 1: ID 2-D objects vs scrambled

Task 2: Reach and grasp 3d objects

(shapes) w/ or w/o pinch

Conclusion: damage to ventral stream;

lateral occipital complex

“What” and “Where” pathways

Conclusion: 2 independent processing streams

Temporal lobe - ventral stream

Perception of objects – “what”

Construct perceptual representation of visual world

Parietal lobe - dorsal stream

Location and action on objects – “where”

Visual control of actions directed at objects

New model by Goodale:

Both streams process structure and spatial attributes of object

Processes are for different purposes: Ventral: object parameters for perception

Dorsal: parameters for control of action

Goodale’s research http://psychology.uwo.ca/faculty/goodale/research/ Two visual systems (dorsal/ventral streams)

Visuomotor psychophysics How are sensory inputs transformed into motor acts

Grasping illusions Deceive eyes but not hand

Virtual grasping Judgments of object size dependent on presence of other objects but

grasping is not (absolute measurement)

Active exploration of objects Actively rotating 3d objects on computer led to faster object

recognition vs passive view group

The effects of retinal motion on reaching

And much more…

Whitney, Westwood, & Goodale (2003). The

influence of visual motion on fast reaching movements

to a stationary object. Nature, 423, 869-873.

Task: Ss focused on bulls-eye; touches screen as quickly as possible where flashing object presented

IV: vertical drifting pattern; upwards or downwards; with unpredictable switch in direction

Conclusion: motion changes reach trajectory

Examine time-course of influence of motion

Perception & action Presentations: Psychonomics Conference 2011

Discrimination of slowed rhythms mimics beat perception impairments in Parkinson’s disease

Action alters object identification: Wielding a gun creates a bias to see guns

Improved visual cognition through stroboscopic training (Mitroff)

Planning for object displacements required walking and reaching (Rosenbaum) How do we decide which paths to take to pick up and carry objects?

What factors are taken into account?

Spatial perception during joint action

Space is perceived from an expanded perspective

Reaching for the star: Rapid reaching influenced by learning reward and probability (Enns) Stimuli that are highly rewarding in a context are associated with

privileged perceptual processing