Title: Is
1Is Non-Conceptual Content Content?
- Louise M. Antony
- The Ohio State University
2Sellarss Problem
- Inspired by Sellarss attack on the given
- Where do (mere) causes stop, and reasons start?
- In McDowells terms, question about possibility
of empirical knowledge
3Empirical Knowledge
- A. McDowell - minimal empiricism the idea that
experience must be a tribunal, mediating the way
our thinking is answerable to the way things
are.
- B. Experience must exist within the space of
reasons there must exist a justificatory
relationship between experience and empirical
beliefs based upon it.
4- (A) requires there to be s.t. wrt which we are
passive, to which we make no contribution, s.t.
the character of which is due entirely to the way
the world is hence, the given - -- since conceptualization is spontaneity, (A)
entails that the given must be unconceptualized
5- (B) requires that experience be brought within
the space of reasons anything else would be
mere exculpation. - -- entails that experience be conceptualized.
- It appears that (A) and (B) cannot both be
satisfied.
6A Naturalistic, Computationalist Translation
- Naturalistic assumption
- Anything relevant to mental processes must be
psychologically tractable - Computationalist assumption
- Psychological tractability computational
tractability
7More computationalist assumptions
- Methodological solipsism
- Computations are sensitive to form and
insensitive to content (except insofar as content
is encoded in form) - Classicism
- Mental processes exploiting rational relations
among thought contents are realized by compl
processes exploiting syntactic relations among
thought vehicles. (Note May be other kinds of
mental processes as well.)
8Normal Perceptual Belief Fixation
- flower ? retina
- retina ? optical nerve (retinal signal)
-
- formation of perceptual belief
something happens
9The Problem, Briefly
- Perceptual belief (4) has intentional content
represents the world as being a certain way. Has
correctness conditions - Retinal signal (RS) simply transduces information
contained in light reflected from flower -- has
no correctness conditions - When does the intentional content appear?
10For perceptual belief to count as knowledge
- There must be a rational relation between RS and
belief - Rational relations hold only among items that
have intentional content - Therefore, IS must have intentional content
11For RS to carry information about the world
- Character of RS should strictly depend upon state
of the world - But if character of RS depends strictly upon
state of the world, RS cannot misrepresent the
world. - Therefore cannot stand in normative relationship
to world. - Therefore, RS cannot have intentional content
12Sellarss Problem Again
- Requirement that RS carry information about the
world (i.e., that RS the given) ? RS cannot be
genuinely representational - Requirement that RS stand in proper epistemic
relationship to perceptual belief ? RS must be
genuinely representational
13A way out?
- Evans, Peacocke, Brewer, Heck, Fodor posit
non-conceptual content - States w/ NCC can
- Serve as reasons for fully conceptualized states
(e.g., perceptual beliefs) - Be faithful registers of information about the
world information isnt packaged into
concepts
14Additional motivations for positing NCC
- Phenomenology of perceptual experience
(richness ineffability arguments) - Animal and infant thought
- Empirical considerations psychl processes
sensitive to information not conceptually
represented.
15What is the distinction?
- Rough idea NCC is picture-like and CC is
language-like - First suggestion (Dretske?)
- Conceptual digital
- Non-conceptual analog
- No Pictures, graphs can be digital in format,
but still represent pictorially
16The Generality Constraint
- If a subject can be credited with the thought
that a is F, then he must have the conceptual
resources for entertaining the thought that a is
G, for every property of being G of which he has
a conception. - Gareth Evans, The
Varieties of
Reference, p. 104 -
17Fodor
- Discursive (conceptual) representation ?
canonical decomposition - Only canonical parts are semantically evaluable
- Iconic (non-conceptual) representation ?
- no canonical decomposition
- Every part is semantically evaluable if R
iconically represents S, then every part of R
represents part of S
18Examples
- The woman on the glacier is playing the flute.
- The woman on the glacier is a constituent is
semantically evaluable - on the glacier is is not a constituent not
semantically evaluable
19Will the NCC proposal work? No
- Empirical side of Sellarss Problem shows that
NCC shouldnt be genuinely intentional - Communication-theoretic considerations
philosophical considerations about nature of
representation show that NCC isnt genuinely
intentional - Genuine intentionality linked to discursiveness
20Solution The Plan
- Communication theory, the nature of
representation, and discursiveness - Unpack knowledge side of Sellarss Problem
-
-
21Grice
- Distinction between natural meaning (meaningN)
and non-natural meaning (meaningNN) - (A) Those spots meantN measles vs.
- (B) The doctors saying measles meantNN measles
- (A) entails that if there are spots, then there
is measles not so with (B).
22Garfield on MeaningN
23Dretske
- A signal r carries the information that s is F
The conditional probability of ss being F, given
r, is 1. - Call a state a Dretskean vehicle if its
occurrence entails the obtaining of an instance
of that type of situation which constitutes the
vehicles informational content
24Gricean vehicles
- For a state to meanNN that s is F, it must be
possible for that state to occur even if it is
not the case that s is F - Call this condition detachability (Antony
Levine, 1991) - To be a Gricean vehicle, a state must be
detachable
25Detachability and the Disjunction Problem
- Detachability is required to distinguish
meaningNN (genuine representation) from meaningN - Disjunction Problem get horse to have the
content horse, even though horse tokens are
sometimes caused by non-horses - Detachability ? solving the disjunction problem
26- Some non-horse-caused tokenings of horse are
mistakes -
- Possibility of mistake correctness conditions
- Having correctness conditions ? detachability
27Dretskean vehicles and the Disjunction Problem
- Information requires no equivocation, but
- whether or not a signal is equivocal depends on
how we carve up possibilities at the source - The informational content of a Dretskean vehicle
is the disjunction of its possible causes --
Dretskean vehicles cannot (must not) solve the
disjunction problem
28- Dretskean vehicles cannot have correctness
conditions - Only Gricean vehicles can have correctness
conditions
29- Detachability, in paradigm cases, comes from
conventionality. Cant be the case with thought
contents. - How to get Gricean vehicles from Dretskean?
- Add information imposition of conceptual
structure makes possible assertion, makes
possible detachability and correctness conditions
30Back to Knowledge Requirement
- Need to rule out cases of mere exculpation
- Bump on head causes belief that Helena is the
capital of Montana (Mere causal process) vs. - Hearing my trusted teacher say Helena is the
capital of Montana causes belief that Helena is
the capital of Montana (rational causal process)
31- Content requirement sufficient to rule out (A) as
epistemically improper but not sufficient to
distinguish (B) from - (C) Bump on the head causes me to believe that I
have a bump on the head
32- (C) Is a case of mere exculpation if the belief
is not based on (Byrne) the experience of
feeling the bump on the head. - Proposal Being based on is a matter of
psychological tractability, hence computational
tractability ? formatting requirement
33- Mere causal processes can be distinguished from
rational causal processes by failure of first to
satisfy formatting requirement dont need
satisfaction of content requirement as well
34- Dretskean vehicles can encode as well as simply
carry information. Encoding a matter of match
between informational format and demands of
computational process
35Intelligible Causal Processes
- Information encoded in a Dretskean vehicle is
s.t., if information were specified discursively,
would provide good evidential basis for
subsequent empirical belief