Perception

Mer som dette

Human Error (ACS)

Ved Darren Smith

Input output devices

Ved Doris McKnight

Boom Crash

Ved umwe bejo

MI

Ved cperez017@ikasle.ehu.eus c

Perception

Visual Illusions

Use of visual illusion can be used for both positions

Knowledge sometimes but not allways involved

So perception depends on the task

Direct perception= no processing

Penrose Triangle

Top-Down:

But its 2D

Hyp: it must be a 3D triangle

Ames room

We know there are no giants

Yet we see one

Object + Environmental clues are enough to create percept

Knowledge is involvedWe produce Perceptional Hypothesis

Most likely one wins

Mask of Hor

Top Down Knowlege

no face, its hollow

Hypo: it must be face

Wrong perception

Strong top down

Necker cube

We constantly build a new perceptual hypo

Kanizsas Triangle

See Triangle although there is none

Strong evidence against direct perception

Marr Detail

4. 2,5 D

Vectors: Distances/ orientation are determined

The Ship

imporverished stimuli are easier recognized one we know what the object is

We can move around

vs. 3D

We can recognize from different angles

Uses Gibson (other modules):

Texture

Stereopsis

Evidence: Motor and Texture processed sepeartly

3. Full Primal

Objects are grouped

Gestalt Principles

2. Raw Primal(shapes)

Points are connected

Termination: Discontinuity

Edge: Change in intensity

Bars (pararllel lines)

Blobs (enclosed arae)

1. Gray level description

Why Blurring

Random fluctionations: Small Changes

Edges: Large changes of intensity

Critique: Colour is important clue

Marr: different module

Marrs information processing

Evaluation:

Enourmously important theory for cog psy

has been implemented

Marr works but doesn't have to be like that

Processing

Algorithms

Mostly Gibsonian with Gestaltist

Gregory

Ames Room: We know there is no giant/dwarf

Makes use of visual illusion

Why?

Ambigous Clues

Incomplete info: Perception under uncertainty

Law of Prägnanz (best gestalt will appear)

Prior knowledge= important

Constantly generate hyptothesis

Top Down

Goal: Perception=for Recognition

Gestaltitst

Most stable representationwill occur: Gestaltfest

Critique

Only explains simple phenomena

Not fully developed: No "How"

Subtopic

Marr: for fore and background

Rules built in

Dorsal Ventral

Evidence for 2 routes

Dorsal (Fast)

Where= Gibson

Drives behaviour

Ventral

What= Gregory

Drives Object recogntion

Patient DF (Milner and Goodale

but grab right size with her fingers

Not recog objects / not know size difference

Normans: Dual Proces Approach

More interconnected than independent

Carlton / Binsted

Learning example

Once Learned: dorsal

More Motor action

Intital Learning: ventral

More conscious processing

Schneiders Hamsters

Pattern recog. vs. orientation in space

Gibson

Direct Perception

How?

Because of poss. ambiguity we Search world for invariant clues

Depth Clues

occlusion: close cows cover up far cows

Relation to Horizon

Far colours more bluish

Faster moving= closer (motion parallax)

Texture Gradients

Vergance to Motor

What?

Evidence

Work with pilots

pilots undistinguishable with 2D Training material

Training with 2D little impact

Evaluation

Critque

Doesn't explain how it works

Cannot explain phenomena

Visual illusions

Neckar Cube

Kanizsas Triangle

Size constancy

Image on the retina gets bigger but doesnt jump at us

We do make assumptions

Need for realistic / non lab investigations

Strong theory: explains a lot

Affordance

No processing

No memory involved

Objects guide action: make evident what can be done with them

Real World has enough clues

Visual Symstem developed in real world

Ecologist: Most info lost in Laboratory Studies

Bottom up

Goal: Perception=for action