Post on 29-Dec-2015
The Nature of Mental Representations
How is knowledge represented in the Mind?
Think carefully about the people depicted.
What do they look like?
Now compare each of their voices
How do these representations exist in your mind?
The Nature of Mental Representation
What does your representation of a penny look like?
One of the greatest problems remaining is understanding the nature of mental representation
Knowledge Structures:Declarative Memory: facts that can be stated (Knowing that)
Procedural knowledge: procedures that can be implemented(Knowing how)
How to study mental representation…*Experiments manipulating represented knowledge*Neuropsychology
Cognitive MapsTolman 1930’s
Experiments by Tolman demonstrated that “representations” exist
• Group 1: Rats had to learn maze, reward for getting from start box to end box was food– Made more errors than group 1
• Group 2: no reinforcement for getting to end box– These rats learned the maze without making errors
• Group 3: no reward for 10 days. On 11th day food placed in end box for first time. – With one reinforcement ran maze as well as group 1
What do Tolman’s Maps look like?
Edward C. Tolman (1886-1959)
Latent learning goes against standard behavioristic principles, which claim that learning comes only from outcomes
Cognitive MapsTolman 1930’s
• Conclusions: Learning a cognitive map: – An internal representation of the maze
Cognitive MapsBees
vonFrisch 1967• Studied behavior of bees returning to hive after
locating nectar• Can form imaginal maps• Can use a symbolic form of communication• Different patterns of dances represent different
meanings
Cognitive MapsBees
vonFrisch 1967• Round dance: source less
than 100 yards from hive• Figure 8 dance: greater
distances
Propositional versus
Analogical Representation
The Age-old Debate
Representational Systems Require:
An orderly mapping between the Represented and Representing worlds
A mental representation is a system of symbols, conscious or unconscious, that are isomorphic to some aspect of the environment, used to make behavior-generating decisions that anticipate events and relations in that environment
C.R. GallistelRutgers University
“Words are the images of things”
The Greek poet Simonides, c. 500 B.C.
How about Book?
“Book” = “Libro”
Theories of Mental Representation
Meaning is grounded in the sensorimotor processes that result from interactions with the world. Semantic representations retain properties of the sensorimotor states that gave rise to them.
Analogical, Modal, & Embodied Meaning:
Meaning is represented by arbitrary codes that do not retain any of the properties of sensorimotor states that gave rise to them.
We don’t store images. Representations resemble abstract form of a preposition. Stored in terms of their deep meanings
Relationship between elements/ Element/ Object
Propositional & Amodal
Propositional
“Hammer” = ?“xuz1245”
Analogical
“Hammer” = ?
Propositional Representations for Circle:
“A series of points all of which are equidistant from a central point”
(Arbitrary code = English, Geometry)
“101101101101101010” (Arbitrary code = Binary Digit)
“x2 + y2 = r2” (Arbitrary code = Analytical Geometry)
Analogical Representation for Circle:
Analogical Representations:
Thermometers
Photographs
Cognitive mapsHumans
• Landmark knowledge: features: imaginal and propositional
• Route road knowledge: specific pathways: procedural and declarative knowledge
• Survey knowledge: estimated distances between landmarks: imaginally or propositionally(number)
Cognitive mapsHumans
Imagery
• Mental representation of things not currently being sensed by the sense organs– Often a recollection of something
• Can imagine things never experienced– Imagine what it would be like to jump from a plane
• Can imagine things that don’t exist– Imagine that Bush wasn’t the current president
• Can use any of the sensory modalities– Most research targets visual imagery
Kosslyn:We use imagery to solve problems
• Which is a darker red: An apple or a cherry?
• How many windows in your bedroom?
• How do you get to the subway from here?
EXPERIMENTS supporting ANALOGICALREPRESENTATIONS
Shephard and Metzler Experiments (1971) Rationale: If mental representations share features with their
referent, that is, they are analogical or isomorphic, thenrotating a mental image should take as much time as rotating
one physically.
Hypothesis: When people rotate a mental image, the farther they have to rotate it, the longer they will take.
Method: Subjects were shown pairs of two-dimensional drawings of three-dimensional objects and were asked whether the drawings in each pair represented the same
object in a different orientation or whether they represented different objects.
Results: 97% Accuracy. The time it took the subjects to provide the answer was proportional to the angular
difference between the two drawings.For example, if the second image was rotated 60 degrees,
the subject would take twice as long than if it was just rotated 30 degrees.
Experiment
• Close eyes and imagine a cube floating in space in front of you
• Now: Grasp the Left Front bottom corner of the cube with your left hand and grasp the right back top corner of the cube with your right hand.
• Now Rotate the cube so that the corner in your left hand is directly below the corner in your right hand
• Describe
Using positron emission tomography (PET), an axial brain slice taken 56 mm above the anterior commissure–posterior commissure (AC–PC) line shows activation in the primary motor cortex (M1) when data from the external-action condition were subtracted from those of the internal-action condition. Depending on the strategy used, motor regions of the brain are recruited during mental rotation. The result also shows that the strategy used to accomplish a given task can vary according to previous training, and can be adopted voluntarily.
Other evidence suggesting that representations are analogical:
1) Mental images of large objects (e.g., an elephant) take longer to scan from left to right than smaller objects
2) Details are more difficult to work out in mental images of small things (fruit fly) than in large objects (airplane). For example, Elephant head to tail last longer than Pig head to Pig tail.Similar with map experiments.
Distance in image is proportional to time spent scanning
Kosslyn 1975
• Imagine :• elephant and a rabbit• Rabbit and a fly• Rabbit and elephant sized fly • Rabbit and a fly sized elephant• TASK: describe details of objects• RESULTS: took longer to describe smaller object
of pair– =Perceptual resolution better for larger objects
Evidence for Analogical Scanning
Stephen Kosslyn (right) and one of his collaborators
• Perception: longer to scan longer distance than short distances
• Experiment to test imaginal scanning– Shown a map of an imaginary island with
landmarks– Studied until could reproduce from memory– Critical phase
• Hear name of an object (tree)• Picture map, mentally scan directly to object and
press a key when have arrived to the object• Hear name of second object, scan to that object’s
location• Measured RT
Image scaling
Findings Distance & RT are Positively Related
Image scaling
• Spatial representations can be scanned in the same way that physical percepts can be scanned
• Our strategies and responses for imaginal scanning are expected to be functionally equivalent to those we use for perceptual scanning
Auditory imaginal representation
• Intons-Peterson et al 1992– Subjects take longer to mentally shift a sound
upward in pitch from the low purring of a cat to a high pitched ringing of a phone than to switch from purr to clock tick.
Zenon Pylyshyn
Problem: Is it all due to task instructions?Demand Characteristics
What is farther west, Reno or San Diego?
Is the mental image of Kosslyn’s island really map-like?
Neuropsychological evidence
• Georgopoulos et al 1989– Single cell recordings in motor cortex of monkeys
– Monkey trained to physically move a handle perpendicular to and counterclockwise to a target light
– During rotation cortical activity recorded
– Later take away handle, cortical activity recorded
– RESULTS: activity in motor cortex showed same cortical cells respond when monkeys are anticipating the movements of the rotations
L.H.Farah’s patient
• Visual imagery(color and shape) distinct from spatial imagery(depth and distance)
• LH head injury at age 18• Lesions in left temporo-occipital regions, right
temporal, right inferior temporal• Anomia- could see but could not apply verbal
label– Bad at describing color and shape imaginal info– Good with spatial manipulations
What processes do we use to represent and manipulate
knowledge?• How we represent info influences how
easily we can perform cognitive tasks
• CMLIX*LVIII
What processes do we use to represent and manipulate
knowledge?• How we represent info influences how
easily we can perform cognitive tasks
• CMLIX*LVIII
• Easier as 959*58
Declarative vs. Procedural
• Declarative: knowing that
• Procedural: knowing how
• Write normal signature as quickly as possible: From first letter of first name to last letter of last name.
• Now turn over paper– Write your signature backward starting with
last letter of last name to first letter of first name
Declarative vs. Procedural
• For both signatures had declarative knowledge about what comes next to each letter
• Only had procedural for the normal writing of name
Concepts
• Idea about something = fundamental unit of symbolic knowledge
• Relates to other concepts– Red, round, fruit
• How do we organize concepts– Categories: organization based on common
features or similarity to a prototype
Schemas
• Can include other schemas
• Encompass typical general facts that can vary from one instance to another
• Vary in their degree of abstraction
• Often have if, then relationships embedded– If a glass falls on to a hard surface, then…
• It should break
Schematic relationships
• Young child calls out “mama!” after he falls from the jungle gym at the playground
• Who is he calling to- a 75 yr old woman, a 45 year old man, and a 30 year old nun or 25 year old woman?– Utilize schemas to answer this questions
Procedural knowledge
• Problem solving• Improved by practice until becomes automatic• Others have made a distinction from declarative
knowledge as “non declarative”– Includes perceptual motor and cognitive skills:
procedural
– Simple associative knowledge: CC
– Priming
Experiment1/2 class close eyes
• ZAZIP
• GASPETHIT
• POCH YUSE
• OWCH MINE
• ILCHI
• ACOT
Other half of class close eyes
• TECKAJ
• STEV
• ASTEREW
• OLACK
• ZELBAR
• ACOT
• PIZZA• SPAGHETTI• CHOP SUEY• CHOW MEIN• CHILI
• TACO OR COAT
• JACKET• VEST• SWEATER• CLOAK• BLAZER
• TACO OR COAT
Declarative/ Procedural differences
• Amnesiacs: can learn new skill but can’t express, can’t consciously remember that they have been learning it.
• Show improvements in
skill tasks: mirror drawing
• One representation is part of non declarative knowledge, trying to make it explicit can be very difficult
Karl S. Lashley and the Problem of Serial Order
Musicians play at a rate so fast that feedback cannot serve as a stimulus for the next note. Pigeons were
taught to learn an arbitrary sequence of colors:
redgreenyellowblue
Successive Chaining paradigm:S(A):R(A) S(B):R(B)S(C):R(C)S(D):R(D)
In successive chaining, each cue is responded to individually
Serial Order & Its Representation
No Possibility of Chaining
Simultaneous Chaining Paradigm:S(A):S(B):S(C):S(D) R(A) R(B)R(C)R(D)
All items are presented simultaneously, and there is no differential feedback during the execution.
Herb Terrace and one of his students at Columbia University
Knowledge of the Ordinal Position of List Items inRhesus Monkeys
Chen, Swartz, & TerracePsychological Science, 1997, pp. 80-86.
LIST1. A1B1C1D12. A2B2C2D23. A3B3C3D4
RESPONSEBirdFlowerFrogShellsTreeWeaselFlyWater
ElkRocksLeavesPerson
NEW LIST1. A2B4C1D32. B3A1D4C23. D1C3B2A4
NEW RESPONSETreeFishFrogPerson
RocksBirdTomatoFlyShellsLeavesWeaselSky
Results: It takes longer for Rhesus monkeys to learn a new list if items are repeated in a different ordinal position.This provides evidence that Rhesus monkeys internally represent ordinal position
Still, one of the greatest problems confronting psychology is the nature of mental representation
http://www.psych.uni.edu/psychexperiments/Exps/labexperiments.htmhttp://www.cog.brown.edu/~tarr/stimuli.html www.sims.berkeley.edu/~sack/ CM/detailed-intro.htm