Spatial Knowledge, Imagery, Visual Memory

closebunkieΤεχνίτη Νοημοσύνη και Ρομποτική

15 Νοε 2013 (πριν από 4 χρόνια και 7 μήνες)

90 εμφανίσεις

Visuospatial Representation

Spatial Knowledge, Imagery,
Visual Memory


What is a representation?

Four aspects of representation

The represented world

The representing world

Set of informational relations on how the
two correspond

Set of processes that extract and use
information from the representation


Mental representations are carriers of

In order to interact appropriately with the
environment we represent info from it and
manipulate those representations


Meaning derived from how representation stands
in consistent relation to the represented world


Meaning determined by relations to other

Spatial Knowledge

How we represent and use spatial

Separate from strictly verbal knowledge

Semantic propositions

Dependent on the linear dimension of

Spatial Cognition

How is the representing world like the
represented world?

The represented world is a space

The representing world is a space

What kinds of processes might be

Space as a representation

Spatial representation

Representing world is a space. What is a space?

Geometric entity in which locations are specified relative to a
set of axes

Dimensionality defined by the number of axes that can point
in independent directions

Of interest is the distance between items, which can be
measured in different ways


Straight line

independent dimensions

Saturation and brightness


Distinct dimensions

Color and size

Space as a representation

Physical world experienced (at least
perceptually) has three dimensions (+ time)

However, the representing world is not
confined to any number of dimensions

Represented world does not need to be

Conceptual info can be represented spatially

More on that later

Spatial Representation

Analog representation

Representation mimics the structure of the
represented world

Multidimensional scaling


Abstract assertions regarding the state of
the represented world

Not tied to a particular sensory modality

Multidimensional Scaling (MDS)


Mathematical technique for taking a set of distances and finding
the best
fitting spatial configuration that corresponds to those

Input: a distance or proximity matrix that describes
how close every object in a set is to every other

N objects are represented by N(N
1)/2 numbers (distances)

Output: a geometric representation where every
object is represented as a point in D

Each object is represented as a point in space

N objects are represented by ND numbers (coordinates)

Purposes of MDS

Give psychological interpretations to the dimensions

Reveal the dimensionality of a data set

Difficult to get a sense of relative distance by means of this information


MDS recovers absolute original locations for the objects from the

Flipping on horizontal axis would give us a rough approximation of

Analog representation


Propositional Representation

(A,B) 10 miles east

(E,C) 20 miles
south, 10 miles east

(F,D) 10 miles
south, 10 miles west

Analog vs. Propositional


Good for configural info

Easy incorporation of new info



Lots of info must be represented

E.g. one point added may require many propositions

Allows for communication of spatial knowledge
and incorporation of additional information not
related to distance

Going south on I35, one must pass through Denton to
get to either Fort Worth or Dallas

Cognitive Maps

Where is Seattle?

Where is Terrill Hall?

Large vs. small
scale space

Hierarchical representation

Small vs. Large
scale space

Maps of small
scale (navigable space)

Cognitive geography

Maps of large
scale space

What is our sense of the locations of items
in the world?

Small scale space

Survey knowledge

Bird’s eye view (map knowledge)

Good for global spatial relations

Easy acquisition

Not so great for orientation

Route knowledge

Gained from navigating through the environment

Locate landmarks and routes within a general frame of

Landmark knowledge

Salient points of reference in the environment

More difficult to acquire but better for navigation in irregular

May lead to survey knowledge

Perhaps a different type

Cognitive collage vs. orientation free

Large scale space

Which is farther north:

Denton, TX or Chicago, IL?

Portland, OR or Portland, ME?

Hierarchical representation of locations

Hierarchical representations

Relative locations of smaller regions are determined
with respect to larger regions.

States are superodinate to cities, countries superordinate to

USA is south of Canada

Maine is just south of Canada

Oregon is well south of Canada

Oregon must be south of Maine

Cities in Oregon must be south of cities in Maine

In this case such cognitive economy works against us

Portland OR is north of Portland ME

Hierarchical representations

Judge relative position
of cities (Stevens and

When superordinate
info congruent with
question, performance

Is x north of y when one
of right side maps

Hierarchical coding

Huttenlocher & Hedges

adjustment model

Combine info across hierarchical levels

If info at subordinate is known with near certainty, there is
no appeal to categorical info

If info at subordinate (fine
grained) levels is at all uncertain,
people use categorical info in estimation

Bias toward center of category

Bayesian approach utilizing prior knowledge

Gist: errors in estimation are due to categorization
rather than nonmetric spatial relations

How are maps learned

From descriptions

Taylor & Tversky: People learned maps from
survey and route descriptions

From navigation

People can assess distance and direction traveled

Integration of information

Visual information

Vestibular information

Maps formed from video games are less accurate
than maps in which people really move

Rotation is particularly important

Using spatial cognition

Adaptive context

Locating and way finding

Tool Use

Mental rotation vs. mental movement

Symbolic representations of space

Drawings, maps, models



Adaptive context

Locating and way finding


Hatchling sea turtles finding the sea

Salmon finding way back home

However these are more behavioral instinct and
imprinting than pure navigation

Desert ant finding direct route home after
meandering paths in featureless environment

Marsh tit stores seeds in holes in a hundred
various places for later retrieval

Locating and way finding

centered systems

centered systems

Hierarchical coding

Locating and way finding

centered system

Location of objects coded relative to self

Updated as we move through the world


Rieser, Guth, Hill (1986)

Participants asked to point out previously learned
locations in unfamiliar room after blindfolded and led
along path

Did not matter whether previously told which location
they’d be asked about, suggesting attentional focus did
not assist in the process

Problem: may not always be accurate over larger
distances without detailed environmental

Locating and way finding

centered system

Object location coded in relation to stable features
of the environment

Requires feature
rich environment providing info
to dominate sense used by organism

If conditions met, then superior to ego

Allows for rechecking of position (no drift from
accumulation of small errors)

Works better for retaining info over long periods of time

Cognitive maps

Both humans and animals display errors in
judgment that cast doubt in positing a true
‘cognitive map’

Animal studies suggest approximation of
distance from a single landmark

Humans make many errors in spatial
judgments that suggest no real metric

Distance from A to B judged different from B to A

Though again this sort of distortion may be
related to categorization (hierarchy)

Tool use

Making, using and designing tools for
interaction with the environment
involves cognitive processes such as
mental rotation and imagery for success

Shephard & Metzler (1971)

Mental rotation vs. Mental Movement

Logically equivalent

However evidence suggests that mental rotation and
taking/mental movement are
psychologically distinct

Selection task

Which these arrays/models would be the correct view from
over there?

Item question

What object would be nearest to you if you were over there?

Specify frame of reference of relative to the observer

Mental rotation vs. Mental Movement

Selection task


Kids (< 10) not so hot at such a task

Usually pick egocentric view

Huttenlocher & Presson

They do much better when asked to do mentally rotate

Can also physically move to new location that matches a
particular array

Suggests conflict between current physically present
perspective and the new (imagined) one they are
trying to obtain

MR allows them to stay put in the physically present room

Physical movement physically transforms that perspective

Mental rotation vs. Mental Movement

Item questions

If kids do not move item questions help (even as
young as 3)

Again, this helps them maintain that egocentric perspective

If asked to mentally rotate, item questions can
actually hurt performance compared to selection

It may be that in item questions, whole array must be
rotated to determine object relations vs a simple ‘rotation’ of
the person or single object in selection task

Gist: mental rotation and mental movement can be
differentially affected depending on the nature of the
question asked, suggesting there may be different
underlying processes involved

Drawings maps and models

Spatial learning from maps differs from learning by
means of navigation

Map learning may aid configural knowledge and allow for
better estimates of distance between points while
navigational learning allows for better route distance
estimation and location of unseen points

Recall survey vs. route knowledge

specific vs. orientation
free learning

Studies show evidence that navigational learning is more a
collection of multiple views than orientation
free, though
may lead to a sort of orientation
free type of knowledge

Sholl & Friedman

Spatial Language

Contrasting experience with communication

Experience spatial relations continuously, but
language is usually discrete (e.g. near vs. far)

Spatial terms function much like other categories
(e.g. fuzzy boundaries, prototypes)

Experience multiple spatial relations
simultaneously, but speak of one relation at a

A frame of reference must be agreed upon in
order to communicate spatial relations

Spatial Language

Despite the difficulties in communicating spatial
knowledge, ambiguities are generally overcome and
information encoded (survey, route knowledge)

However it does seem that spatial language may bias
or constrain spatial representation, and may even
affect the development of spatial concepts and

Even so, the actual link between spatial language
and spatial representation is not entirely clear

Impaired sight individuals may have difficulties with a
variety of spatial tasks but have intact spatial language


Spatial cognition also contributes to logical
reasoning, metaphor, and creativity

Transitive reasoning

A > B, B > C

A ? C


The future stretched in front of them

My heart is a flame turned upside down

Structural alignment of spatial and temporal

Diagrams as aids to understanding

Show conceptual similarity of items, connections
amongst various concepts etc.


E.g. visualization for problem solving

Taking someone else’s point of view?


Some information in memory is purely verbal

Who wrote the Gettysburg address?

Other memories seem to involve mental

Trying to recall a procedure

Making novel comparisons of visual items

What is a mental image?

How are mental images represented and

Are mental images like visual images?

Evidence for use of visual imagery

Selective interference

Segal & Fusella

Imagery interferes with detection of stimuli
(sensitivity decreased)

Auditory imagery interfered with auditory
detection, visual imagery with visual stimuli

Manipulation of images

Mental rotation studies

Evidence for use of visual imagery


Learn a map

Mentally travel from
one point to another

Measure time to
make this mental trip


Time to make trip
increases with

Times increase with
imagined size of the

Evidence for use of visual imagery

Moyer 1973

Subjects were given the
names of two common
animals and asked to judge
which was larger

Which is larger, a moose or
a roach?

Wolf or Lion?

The time delays as a
function of size difference
were similar to those usually
found for perceptual


Kosslyn 1975

Scenario I: Imagine an elephant
standing next to a rabbit. Does a
rabbit have a beak?

Scenario II: Imagine a fly standing
next to a rabbit. Does a rabbit have
an eyebrow?

People made faster judgments when
relying on a larger mental image
(such as the rabbit next to the fly)
than when using a smaller mental
image (such as the rabbit next to an

Kosslyn suggested that the size of
an image is an important factor in
determining how fast we can make
judgments about it.









Paivio's Dual
Coding Theory

Information is mentally represented either
in a verbal system (propositional) or a
nonverbal (analogical) system (or both).

Each system contains different kinds of

Each concept is connected to other
related concepts in the same system and
the other system.

Activating any one concept also leads to
activation of closely related concepts.


The hypothesis of multiple codes
(verbal and spatial) is based on the
demonstration of independence of

Pictures of objects

Words of objects

Paivio (1975) compared reaction times for


visual stimuli

If the stimuli are processed semantically, there
should be no difference between consistent and
inconsistent presentations.

If stimuli are processed spatially, inconsistent stimuli
should require a mental conversion to appropriate

Which takes time




“Which is larger?”














Congruity Effect only for Pictures (not words)

Imagery relies on perceptual detail and semantic does not

Such findings as this and picture superiority effect (pictures are
better recognized than words), and that verbal + imagery
encoding leads to best recall, suggest a Dual Code Theory







Santa 1977

More evidence of dual

Ss presented array of
objects or words

On test presentation
asked whether the
elements were same as

E.g. In geometric
condition first two would
be yes responses

Santa 1977

Results of positive

Spatial configuration is
preserved in geometric

Compared to verbal
presentation, which was
encoding in typical
English reading style
and benefited from the
linear configuration

Representation of images

What is the relationship between
imagery and perception?

Can imagining interfere or facilitate
detection of stimuli?

Similar processes involved?

Spatial Knowledge

Symbolic Distance Effect (Moyer 1973)

Process of imagery = process of perception

As perceptual distance increases so does
psychological distance (RT).

Items “farther apart” are more quickly distinguished

Which is larger?



Representation of images

Contrary evidence

Chambers and Reisberg

Images are (committed
to) a particular

E.g. The rabbit comes
once drawn but was only
a duck as imaged

Contrast with
perception which
requires interpretation,
images are already

Are visual images visual?

Plenty of evidence to suggest a spatial
component to visual imagery, but perhaps
the visual part is represented propositionally


Congenitally blind also take longer to imagine
longer map routes like the one in Kosslyn

Are visual images visual?

Images are also not as sharp as real

Form a mental image of a tiger

Does it have stripes?

How many?

It is hard to examine details of mental
images that would require eye

Making new pictures

Finke, Pinker, Farah


Imagine a capital letter H and a triangle

Rotate the H 90 degrees

Place the triangle on top of it

What is it?

Suggests images can take on new

Are visual images visual?

Facilitation and interference (Farah)

Have people imagine a letter (H or T)

Present one of the letters to the screen briefly (20
ms), or present nothing, followed by mask

Asked if they saw a letter

People are more likely to detect the stimulus
if it was the same as what they were
imaging, suggesting that visual and imaginal
representations joined or fed into the same

Are visual images visual?

Evidence from neuroscience

Patients with lesions of
visual cortex that lead to
perceptual problems also
have problems with mental

ERP evidence PET evidence:
Visual imagery leads to
activation of visual cortex.
Auditory imagery does not

In general, results of studies
from mental rotation to brain
imaging support the idea of
both visual and spatial
representation of images

Translating Words to images

Franklin and Tversky

Create a mental image based
on the description

Asked to identify location of
items in that imagined

Results are what one might
expect given an imagined
spatial environment

down, front
back more relevant
in navigating real world

right confusion in real world
and imagined world

Visual memory

Although our visual
memory seems to be
excellent, it turns out
not to be that great in
many respects

In general, our memory
for details is lost, much
like with other types of

Visual memory

Memory for pictures is quite good

Again, don’t get too detailed


Presented 10000 photos over several days

New memory over 80%

Picture superiority effect

Better memory for pictures than words