Lecture 2 Selecting: Features, Places or Objects? The role of selection in the mind-world connection

1
Lecture 2Selecting Features, Places or
Objects?The role of selection in the mind-world
connection
2
Outline of Lecture 2 Selection Objects or
Feature-Placing?

• 1 Selection The role of selective attention
• 1.1 Allocating and shifting attention The role
  of objects vs places
• 1.2 Studies in object-based attention
• 2 More on what is selected by FINSTs
• 2.1 Causes and codes
• 2.2 Conceptual, nonconceptual, and
  quasi-representational contents
• 3 The relevance of this research to understanding
  sentience ( sensory experience of the world)
• 3.1 Austen Clark and Feature-Placing in
  Sentience
• 3.1.1 Feature placing and the binding problem
• 3.1.2 Feature-placing and the causal link
• 3.1.3 Feature-placing and nonconceptual
  selection
• 3.2 Reprise on FINST theory
• 4 Summary

3
The functions of focal attention

• A central idea in this lecture will be the notion
  of picking out or selecting. The usual
  mechanism that is invoked in explaining selection
  is focal attention.
• Why must we select anyway? There are several
  reasons
• We need to select because we cant process all
  the information available. This is the
  resource-limitation reason.
• We need to select because certain patterns cannot
  be computed without marking certain elements of a
  scene (e.g., in counting)
• We need to select because of the way relevant
  information in the world is packaged (Strawsons
  Collecting Principle). Selection is needed to
  solve the Binding Problem
• Certain kinds of attentional selections
  constitute the most primitive causal contact
  between world and mind. Because predicating
  applies to things in the world, binding predicate
  arguments to things precedes all conceptualizing

4
What does visual attention select?

• If attention is selection, what does visual
  attention select?
• One possible answer is places. We can select
  places by moving our eyes so our gaze lands on
  different places.
• Must we always move our eyes to change what we
  attend?
• Studies of Covert Attention-Movement Posner
  (1980)
• How does attention switch from one thing to
  another?
• Automatic (exogenous) vs voluntary (endogenous)
  selection
• Sperlings episodoc theory of attention
  allocation
• Can attention select empty places?

5
Exogenous (automatic) movements of attention
Example of an experiment using a cue-validity
paradigm for showing that the locus of attention
moves without eye movements and for estimating
its speed. Posner, M. I. (1980). Orienting of
Attention. Quarterly Journal of Experimental
Psychology, 32, 3-25.
6
Endogenous (voluntary) movements of attention
7
The difference between automatic and voluntary
control of attention

• Exogenous and endogenous control differ in
  several ways
• Automatic attention shifts are faster and the
  effects are stronger.
• Voluntary attention shifts can be interrupted by
  exogenous cues, so it is considered secondary to
  automatic control
• With voluntary attention the person only knows
  which direction to move attention, so it may
  occupy intermediate locations
• With automatic shift the apparent detection
  improvement at intermediate locations can be
  explained by decreasing attention at the source
  and increasing attention at the target (Sperling
  Weichselgarter, 1995).
• It is doubtful that there is attentional
  selection of empty regions empty space does not
  have the causal power to attract exogenous
  attention and there is also some doubt that
  voluntary movements can be continuous Pylyshyn
  Cohen, 1999.

8
Evidence that attention is object-based

• Although the earliest evidence (Posner) suggested
  that attention moves through space without eye
  movments (called covert movement) there is now
  evidence that attention attaches to objects as
  a whole
• The main source of evidence was based on single
  object superiority effect
• But now there is evidence from moving objects
  (e.g., MOT, IOR, Object-specific priming)

9
Single object superiority even when the shapes
are controlled
Pay attention to the blue object(s). Which
vertex is higher, the left or the right? Pay
attention to the red object(s). Which vertex is
higher, the left or the right?
10
Attention spreads over perceived objects
Spreads to B and not C
Spreads to C and not B
Spreads to B and not C
Spreads to C and not B
Using a priming method (Egly, Driver Rafal,
1994) showed that the effect of a prime spreads
to other parts of the same visual object compared
to equally distant parts of different objects.
11
Inhibition of return

• Inhibition-of-return is the phenomenon whereby an
  object that has been attended is less likely to
  attract attention again in a period of 300 ms to
  900 ms after it is first attended. The attended
  item is said to be inhibited.
• This is thought to help in visual search since it
  prevents previously visited objects from being
  revisited
• IOR is Object-Based (the only counter-evidence
  involves easily-marked locations like between
  two objects)

12
But IOR appears to be object-based (so it travels
with the object that was attended)
13
Objects endure despite changes in location and
they carry their history with them
object-specific priming
Object File Theory of Kahneman Treisman
Letters are faster to read if they appear in the
same box where they appeared initially. Priming
travels with the object. According to the
theory, when an object first appears, a file is
created for it and the properties of the object
are encoded and subsequently accessed through
this object-file.
14
Visual neglect syndrome is object-based
When a right neglect patient is shown a dumbbell
that rotates, the patient continues to neglect
the object that had been on the right, even
though It is now on the left (Behrmann Tipper,
1999).
15
Simultanagnosic (Balint Syndrome) patients appear
to attend to only one thing at a time
Simultanagnosic patients cannot judge the
relative length of two lines, but they can tell
that a figure made by connecting the ends of the
lines is not a rectangle but a trapezoid (Holmes
Horax, 1919).
16
Balint patients can only attend to one object at
a time even if they are overlapping
Luria, 1959
17
The Binding Problem

• The visual system must distinguish between sets
  of properties present somewhere in a scene from
  sets of properties belonging to a particular
  individual object ?
• Distinguishing properties belonging to one
  particular individual is called the Binding
  Problem. Treisman proposed that binding features
  together requires focal attention and feature
  maps to cross-reference locations
• Treismans Feature Integration Theory and Clarks
  theory of sentience both assume that the binding
  problem is solved by representing the
  spatio-temporal location of properties so
  selection of space-time regions is required in
  order to group properties appropriately. This is
  equivalent to the assumption that sensation must
  be expressible in a feature-placing language
  (Strawson)

18
Selfriges Pandemonium feature-based vision
system does not address the binding problem
19
Sketch of a possible neural network
implementation of FINSTs
20
Details of the Winner-take-all FINST network
21
Some further distinctions we must make to
understand nonconceptual representation

• Distinguish causes and codes
• What causes Object Files to be created vs what
  conceptual information is entered into them (cf
  Kripkes distinction between what something
  refers to and the properties that fix the
  reference)
• Conceptual, nonconceptual and quasi-representation
  al contents
• Representing vs carrying-information-about
• The case of clusters, figure-ground, and
  correspondence
• How information-carrying properties (e.g.,
  location on the proximal pattern) can create
  clusters without, in an important sense,
  representing the distal location of the features
  that are clustered
• Aside on the special status of analogue
  representation

22
Austen Clark ( P. Strawson) on feature-placing
languages

• What kind of representation does sensation
  require?
• Ans Just those expressable in feature-placing
  languages

The hypothesis that this book offers is that
sensation is feature-placing a pre-linguistic
system of mental representation. Mechanisms of
spatio-temporal discrimination serve to pick
out or identify the subject-matter of sensory
representation. That subject-matter turns out
invariably to be some place-time in or around the
body of the sentient organism. The aboutness
of sensation reduces to its spatial
character. there is a sensory level of
identification of place-times that is more
primitive than the identification of
three-dimensional material objects. Below our
conceptual scheme underneath the streets, so to
speak we find evidence of this more primitive
system. The sensory identification of
place-times is independent of the identification
of objects one can place features even though
one lacks the latter conceptual scheme. Clark,
A. (2000). A Theory of Sentience. Oxford. (p
165)
23
The relevance of the present research to the
problem of sentience

• Austen Clark and Feature Placing
• Spatiotemporal regions and the binding problem
• Spatiotemporal regions and the causal link
• Spatiotemporal regions and nonconceptual
  reference
• Where does FINST theory fit in?

24
Problems with the location-based solution

• Location-based views of how binding is carried
  out have several fatal flaws
• Choosing the size and shape of the relevant
  selected region presupposes that the object has
  been identified ?
• Sensation (i.e., early vision) requires that the
  world exert a causal effect upon vision. Empty
  spaces have no causal effect (they have no
  intrinsic properties).
• Nonconceptual selection is a direct, unmediated
  selection caused by events in the world.
  Whatever is selected is not selected under any
  concept or encoding not location, not object,
  not anything!
• So the selection could in principle be a
  property, a place, a region, or an object, but it
  cannot be encoded as such!

25
Vision can distinguish objects that differ by
conjunctions of properties, so early vision must
not lose the object-specificity of properties it
detects. Can it do this be selecting a region
without regard to the object that occupies that
region?
26
Problems with the location-based solution

• Location-based views of how binding is carried
  out have several fatal flaws
• Choosing the size and shape of the relevant
  selected region presupposes that the object has
  been identified ?
• Sensation (i.e., early vision) requires that the
  world exert a causal effect (an interrupt) upon
  vision. Empty spaces have no causal effect (they
  have no intrinsic properties).
• Nonconceptual selection is a direct, unmediated
  selection caused by events in the world.
  Whatever is selected is not selected under any
  concept or encoding not location, not object,
  not anything!
• So the selection could in principle be a
  property, a place, a region, or an object, but it
  cannot be encoded as such!

27
Where do we stand?
28
(No Transcript)
29
FINSTs and nonconceptual representation (a
reprise)

• What does the early (nonconceptual) vision system
  deliver to the mind?
• What classes and properties can be recognized
  without the apparatus of concepts?
• Causality? Cardinality (of small sets)?
• 3D object shapes? Shape-from motion? Shape from
  shading? Shape from contours?
• What, after all this, can be selected in a
  nonconceptual manner, and how does this help with
  the problem of connecting vision with the world?

30
Going beyond nonconceptual representations

• Work with Infantss numerosity judgment
  frequently appeals to Index theory (Leslie,
  Carey, Wynne,)
• Some of these findings appear to implicate
  indexing of nonconceptual properties, but some
  may not the distinction is not easy to draw in
  practice e,g.
• Infants can use certain properties to decide to
  create an new object file but not to recognize if
  an object is the same one that caused the object
  file to be created earlier
• Determination of cardinality in infants appears
  to be sensitive to such properties of individuals
  as whether they can be taken apart, whether they
  were poured, whether their parts moved together,
  etc.
• Such effects may indicate either that infants are
  deploying concepts or it may indicate that the
  mode of arrival of individuals affects whether
  they are indexed and tracked. This appears to be
  true of adult tracking as well!