Psy 348 Fall 2008:  Knowledge representation & visual imagery

Two types of knowledge

            Declarative = “Knowing that . . .” [Gilbert Ryle]

                        Examples

            Procedural = “Knowing how to . . .” 

                        Examples

            These different types may generate different kinds of mental representations

Two major theoretical positions regarding how knowledge is represented mentally – Dual Code versus Propositions

            Dual Code – Allan Pavio, Stephen Kosslyn

                        Some info is represented verbally

                        Abstract concepts are especially strong here:  love, justice, democracy . . .

                        Verbal info is usually represented symbolically – the representations are

                                     arbitrary just like most words, numerals

                                      Examples . . .

                        Some info is represented pictorially

                                    Concrete objects and geometric shapes are notable here . . .

This representation is often called “analogue” because it is analogous to

the real-world object that it represents

                                                Examples . . .

                                    Mental imagery – Kosslyn (vision – image scaling & scanning);

                                                Intons-Peterson (auditory); Farah (neuro);

                                                Finke, et al (combining images);

                                                --representing things not currently being sensed (i.e., memory)

                                                --can represent information in any sensory modality

                                                Examples . . .

                                    Mental rotation – Shepard & Metzler

Form a mental image of an object, then imagine that is it rotated to a different position;

“rotating” the imagined object takes time, just like rotating a real object.

Hence, the functional equivalence hypothesis –

mental representations are functionally equivalent to real objects

                                                Examples . . .

                        Summary:  Two codes, one verbal and one imaginal

                                    A compelling theoretical position, with lots of empirical support

                                    Nevertheless not all of the characteristics of the referent are retained

                                    Furthermore, western psychologists insist that there are strong limits to how complex an

                                        image can be [but contrast this to Buddhist meditation theory -- see below, "Applications"]

                                    How to account for discrepancies & deficiencies?

           
            Propositions – Anderson & Bower, Pylyshyn, Chambers

                        Knowledge is represented in abstract forms similar to predicate calculus:
                                    Relationship [Subject element, Object element]

                                                Examples:  “Cat is under the table” = Under [cat, table]

                                                “The black cat is under the round table” =

                                                Under [cat(black), table(round)]

                                    No pictures or words are involved

                                    The sensation that we have mental pictures or verbal thoughts is simply

                                                an “epiphenomenon” of propositional processes

                                    Possible advantage is that propositions can represent any kind of

Relation and any combination of relations,

including very complex ones

                                    Problem:  how do we go from a set of abstracted propositions back to

the “concrete” verbal or imaginal code? 

                                    Answer:  mental processes “recreate” the verbal or imaginal code.

                                                Problem: How? And what mental processes? 

Is there a homunculus lurking here?

                                    Aside re: homunculus

            Synthesis – Tversky and others

                        We can “create” mental maps from verbal descriptions

                        We can create images of maps we’ve actually seen

                        Accuracies & distortions are similar for both of the foregoing

                        So perhaps we use both dual code & propositions

Applications

            Memory for lists

                        Interacting images

                        Method of loci

                        Pegword technique

            Solving problems

            Athletic/sports training

            Non-verbal knowledge via meditation (especially in Buddhist traditions)               

General summary:

            So three possible codes: verbal, pictorial, propositional

            Dual code / functional equivalence theories have overwhelming empirical research support.

Propositional theories tend to rely on rational arguments & less empirical research.

            Currently the two positions are usually portrayed as mutually exclusive: 

                        one is right and the other is wrong.

            Most cognitive psychologists currently go with dual code / functional equivalence.

            Barbara Tversky & others:  maybe the two positions are not exclusive, but rather

                        we use propositions in some circumstances & dual code in others.

            Visual imagery is often used in memory tasks, problem-solving, skills training, and meditation.