Representations

1
Representations
2

• vast quantity of on-going research dependent
  on Rs on R, no scientist knows how mental Rs
  represent.
• Representing - relation two things one stands in
  for other. stand-in R (wordspictures)
• Mental Rs - implemented in brain
• Implementation Inverse of supervenience Minds
  supervene on brains, brains implement minds
  (Dietrich 2007)
• Supervenience No mental change without
  physical change

3

• Wheleer 05
• Duality mental-physical pp. duality
  subject-object
• Pr. I of Cartesian psychology (Pr of R) Mind,
  cognition, and intelligence explained in Rs
  (manipulated/transformed)
• Pr. II Rs and subjectobject dichotomy
  interconnected
• Rs are context independent

4

• Principle of explanatory disembodiment
• (Cartesian psychology)
• Informational contents carried by bodily
  sensations primitive perceptual states -
  specified in bodily states ? Reliable and
  flexible intelligent action No agents physical
  embodiment off-line cognition
• Homuncularity is necessary for R

5

• Contents and vehicles of Rs
• Contents (meaning) Semantic value of a R
• Contents determined by object in world, the
  relation among Rs or both
• Vehicles Syntactic structures or physical
  realisation of objects that play role of Rs
  (carrying a content) (Eliasmith 07)
• Contents What Rs are Rs of
• Vehicles Rs themselves (Mandik 2003)

6

• Format of Rs
• Propositional (Fodor, Pylyshyn)
• Image (Metzler, Shepard, Kosslyn) (ex picture,
  maps)
• Analogous (Mandler, Lakoff, Fauconnier)
• R Stored information or stands-in for something
• Function Carry information
• Conceptual level - Symbolical Rs Static
  discrete ? Computations

7

• Fodor/Pylyshyn - Correspondence Rs and neural
  patterns - not univocal (Fodor and Pylyshyn 1988)
• Subconceptual level (connectionism) -
  Subsymbolic or distributed Rs - Content depend
  upon processing network environment
• Fodor,Pylyshyn Neural level Implemen. of
  conceptual Rs
• Dynamical ST No Rs/Computations

8

• Debates R on format or nature
• Fodor/Pylyshyn
• Compositionality
• Productivity
• Systematicity
• Smolensky (88)
• Implicit functional compositionality assure the
  advantages of a compositional structure to the
  brain

9

• Dietrich and Markman ('03)
• Cognition Must use discrete Rs
• Discrete Rs A system have discrete Rs iff
  contains more than one R (discrete and static)
  categorize inputs enviroment ? Semantic memory
• Discrete Rs Contains more than one R, and Rs -
  bounded and uniquely identifiable ('07)
• Computationalism vs. dynamical systems approach ?
  Difference discrete-continuous Rs

10

• 7 reasons for Discrete Rs A cognitive system
  must
• Discriminate among states in represented world
• Acces specific properties of Rs
• Combine Rs
• Cog systems combinatorial structure
• Functional role among concepts
• Have abstractions
• Have non-nomic Rs

11

• Image Rs
• Shepard and Metzler (1971) Kosslyn and Shwartz
  (1977)
• Skeptical (Pylyshyn 1984, 2002)
• Numerous computational, psychological, and
  neurological pictures words

12
?
13

• Verbal description information man in chair
  is close to man with his head open, who is on a
  couch
• Verbally infer that chair is close to couch.
• Using pictorial R, however, no inference is
  necessary we can just see that

14

• Visual Rs - accessible to different kinds of
  computations than verbal Rs
• 1. Inspect
• 2. Find
• 3. Zoom
• 4. Rotate
• 5. Transform
• Finke, Pinker, and Farah (1989) Imagine the
  letter Y. Put a small circle at bottom of it.
  Add a horizontal line halfway up. Now rotate
  figure 180 degrees. (Thagard 05)

15

• Kossylyn Image Rs (IRs)
• fMRI
• Wraga and Kossylyn (2003) MI, 2 meanings
  Seeing with mind eyes internal Rs ?
  Perception without sensory input
• Mental imageries Transformation/manipulation
• 1) Scanning (Subject memorizing landmarks from a
  map faster away targets from initial landmark,
  longer took subjects to locate them ? IRs are
  depictive Preserve spatial extend)

16

• 2) Zooming (to see fine detail from an imaged
  object subject can zoom it - farther away the
  object appears in image, longer it takes to
  report correct answer)
• 3) Rotation (Shepard et all. - Two multiarmed
  objects, task being if those objects are same or
  not ? Subjects mentally rotated one object into
  alignment with other)
• Conclusion Internal processes - analogous to
  physical processes

17

• Damage to occipital lobe impairs visual imagery.
• fMRI Use of visual mental imagery - same brain
  areas in visual perception
• Imagery relies on regions of cortex
• that are spatially organized in ways that
  correspond to structure of retina, networks of
  nerve cells that send impulses to brain. (Thagard
  05, p. 106)

18

• Areas of brain immediately connected to retina
  spatial organisation Structurally similar to
  that of retina
• ? Preserve some spatial structure of objects
  presented to retina, their activation during
  imagery suggests that imagery involves
  picture-like Rs
• Behavioral brain-imaging experts Mathematical
  intuition sometimes depends on visual and spatial
  Rs (Thagard 05)

19

• Pylyshyn Propositional Rs (PRs)
• Tacit knowledge of physical laws and objects due
  to previous experience - Less time to rotate a
  longer distance than a shorter one
• This knowledge - stored in mind as abstract or
  language-like form
• Today accepted - 2 forms of Rs (Pavio et all
  Imagery and verbal processes, 1971)

20

• Fodor (2008)
• Conceptual vs. nonconceptual Rs
• Propozitional vs image Rs or
• Discursive vs. Iconic (related to pictorial and
  continuous)
• Nature of iconic Rs - Not conceptual
• a R Compositional iff its syntactic structure
  semantic content - determined by syntactic
  structure and semantic content of its constituent
  parts (p. 171)

21

• Discursive R
• Every sentence Finite arrangement of
• constituents that are themselves either
  primitive or complex
• Each complex constituent is a finite arrangement
  of lexical primitives (words).
• Lexical primitives have their syntactic and
• semantic properties intrinsically

22

• Semantic interpretation of a sentence (of any
  discursive representation) depends on way that
  properties of its lexical primitives
• interact with properties of its constituent
  structure.

23

• Iconic Rs (p. 173)
• Picture principle If P a picture of X, then
  parts of P are pictures of parts of X
• Pictures differ from sentences icons
• dont have canonical decompositions into parts
  all the parts of an icon are ipso facto
  constituents (ex. Picture of a person)
• NO canonical decomposition
• NO constituent structure
• Homogeneous (syntactic semantic)

24

• Discursive Rs (decompose syntactically and
  semantically heterogeneous constituents) ?
  Logical forms
• Iconic Rs No decomposition ? No logical forms ?
  Iconic symbols cant represent things vs.
  discursive symbols can
• Ex Propositions - Distinction between
  negative-affirmative, quantified hypothetical
  modal
• Pictures dont have truth-conditions

25

• Binding problem (B)
• Visual information - divided into separate
  processing streams in retina and LGN, but
  reintegrated later (Horst 07, p. 166)
• We perceive one object (color, shape, etc)
• but
• Separate neural areas for color, form, motion.

26

• Teoria de integrare a trasaturii (Treisman anii
  80 90)
• Milner (1974), von der Malsburg (1981), Singer
  (90, 00) Binding-by-synchrony
• Singer (2007) - Primele experimente
• Coordonarea - prin sincronizarea activitatii
  diferitelor patternuri neuronale (patternurile au
  aceiasi faza a oscilatiilor frecventa fiind de
  40 Hz) ? O singura R a obiectului