Empiricism and the Philosophy of Mind 1

1
Empiricism and the Philosophy of Mind 1

• Dispensing with the given sense data first.

2
The project

• Sellars aim here is to sort out his philosophy
  of mind while retaining an empiricist
  epistemology.
• Issues
• The given Sellars rejects the notion that
  anything is immediately apprehended in propria
  persona there is no transparent awareness of
  anything.
• Sense-data talk and the relation between our use
  of seems and ordinary observation reports.
• How induction can be used to justify our claims
  (as individuals) to be reliable observers.
• Relations between rationalism and empiricism in
  our philosophical tradition. Sellars wants to
  reconcile these, recognizing the importance of
  empirical evidence while also recognizing the
  crucial role concepts play in all knowledge.

3
The Given

• So what is the given?
• Not just something that we observe directly, i.e.
  a claim we accept without inferring it from other
  claims weve accepted.
• Many things have been said to be given sense
  contents, material objects, universals,
  propositions, real connections, first principles,
  even givenness itself.
• A framework that is part of many different
  philosophical positions.
• Among the first forms to be criticized were
  intuited first principles, synthetic necessary
  connections.
• Sellars aims to root it out entirely, not just
  some particular forms of it.

4
Why attack the given?

• Cognitive psychology our accounts of what and
  how we know need to fit with our natural
  understanding of what humans can learn to detect
  and discriminate.
• The neo-Hobbesian picture of humans that ends
  Phenomenalism is incompatible with a view of
  knowledge involving an immediate link between a
  logically simple subject and some fact about the
  world.
• All knowledge is expressed in language, and
  language can shift and change in many ways.
• Our relation to the community and the conceptual
  capacities that arise through language
  instruction are indispensable elements in any
  knowledge we have, no matter how direct (i.e.
  non-inferential) it is.

5
First target Sense data

• For x to be sensed is for it to be the object of
  an act.
• A sense content is a possible object for such an
  act.
• Kinds of sensing may just reduce to the kinds of
  sense content that are their objects.
• The point of these acts is to serve as a starting
  point (foundation) for empirical knowledge.
• But sense contents are particulars, the objects
  of some sensings (i.e. acts of sensing).
• And knowledge is about facts, not particulars.
  (Broadly, it needs a subject-predicate form
  That this object is characterized in some
  particular way.)

6
The dilemma

• The SD theorist has to choose between
• Sense contents are particulars, and sensing is
  not knowing.
• Sensing is a form of knowing facts, not
  particulars, are sensed.
• Even if she adopts the first, she can still say
  that sensory knowledge is logically dependent on
  sensings.
• But SD theorists typically want to have their
  cake and eat it too.
• This can be done, with a little slippery move
• To sense a sense content is to sense it as having
  a certain character.
• When we sense a content as having a certain
  character, we know that it has that character.
• When we sense a sense content, we can say now
  that we know it, without specifying the character
  we sense it as having.

7
Acquaintance

• This line gets comfort from the familiar fact
  that we often do speak of knowing particulars, as
  in I know Fred.
• But the proposal before us links this to
  non-inferential knowledge by the brute force of a
  contextual definition To know x (where x is a
  sense content) just is to know that x has some
  character.
• Sometimes this is ignored in accounts of the
  givenness of sense contents but if sensings are
  analyzable, we can recover the link to knowledge
  if the analysis matches that of non-inferential
  knowledge of some sensed fact about the sense
  content.

8
The given part

• Descriptive accounts of sense contents have to
  smuggle this link back inAgain, Sellars rejects
  any reduction of the normative to the
  descriptive.
• The main point here is that the sensing of sense
  contents is not taken to require any learning,
  setting up of associations, etc. It is a
  primitive epistemic capacity built into us at the
  outset.
• As primitive, it is the starting point for the
  rest of our knowledge Knowledge is acquired
  first in this form, and then extended to wider
  sorts of knowledge

9
But categorization is acquired

• Theres a strong tendency in the field to regard
  knowledge that x is Y requires learning to tell
  Ys from non-Ys, the acquisition of the concept
  of a Y, perhaps even the use of symbols for Yness
  other components of this bit of knowledge.
• The triad
• A. x senses red sense content s entails x
  non-inferentially knows s is red.
• B. The ability to sense sense contents is
  unacquired.
• C. The ability to know facts of the form x is ?
  is acquired.

10
The SD theorists choices

• Give up the first (A). Sensings now can be
  involved in knowings, but they arent really
  knowledge themselves.
• Give up the second. Now sensings arent really an
  account of sensations, which we can have without
  learning
• Give up the third. This is to give up the
  nominalist proclivities of the empiricist
  tradition.

11
The diagnosis

• Sensings look to be a mongrel concept combining
  ideas about sensations (states that play a
  critical role in sensory knowledge and are
  possible states for us from the outset, but are
  not knowings in themselves) with ideas about
  non-inferential knowings that are at the heart of
  our empirical knowledge about the world around
  us.

12
Sources of the confusion

• A scientific explanation of the facts of
  perception (and misperception) justifies the
  postulation of inner states that are normally
  brought about by objects with various sensible
  characters under normal conditions. But there is
  nothing epistemic about such states their role,
  thus far, is causal.

13
Second source

• The argument from security If our knowledge of
  particular matters of fact starts with the things
  we normally report as observations (this is red,
  thats triangular, etc.) then it starts with
  commitments some of which are false there is no
  mark we can introspect that separates the
  veridical ones from the non-veridical. This
  seems to undermine the very idea of empirical
  knowledge.

14
The slide

• looks or seems talk has an apparent advantage
  here. If we want certainty, we can retreat (by
  degrees), getting
• There is an object over there thats red and
  triangular on the facing side.
• There is an object over there that looks red and
  triangulary on the facing side.
• There looks to be an object over there thats red
  and triangular on the facing side.
• Each is increasingly secure the last seems
  almost perfectly so.

15
Describing the inner states

• Now it becomes tempting to take the third of
  these as a description of ones inner state (its
  the sort of state normally produced by objects
  that are red and triangular on the facing side).
• And this state is harder to be fooled about (its
  more intimately linked to our faculty of
  judgment than the properties of an external
  object).
• Further, it just doesnt make sense to talk about
  an unveridical sensation.
• Heres the key mis-step This implies that they
  arent part of the space of reasons at all (they
  are causal- we simply have sensations, i.e. they
  occur in us and that all). Making them the
  foundation of knowledge takes a position, rather
  than report what is given to us.

16
The sense-datum code

• If we introduce sense datum talk as a code for
  seems talk, we need to be very careful not to
  treat the code as though it were a language
  with internal logical structure that we can
  unpack to identify theoretical commitments.
• The inferences we make in a sense-datum code must
  follow those that are allowed when the code is
  translated back into the language it is code for.
• Theories are not just codes they are proper
  languages, which add content to the assertions
  they are linked to by bridge rules.

17
A successful code

• If the illuminates our understanding of knowledge
  and ordinary things, its because it leads us to
  recognize that in fact we can trade in talk of
  ordinary objects for seems talk in its most
  non-commital sense There looks to be
• Then ordinary things turn out to be
  constructions out of lookings or appearings.
• But this is untenable, says Sellars

18
The Logic of looks

• Is looks a relation (between the subject, a
  thing and a quality in x looks ? to s)?
• Broad something elliptical must be before our
  minds when a round penny looks elliptical.
  Sense data (which have the characters things look
  to have) are an explanation of such appearings.
• Some resist such analyses, though, saying that
  there need be nothing red involved when something
  looks red to someone.
• Sellars starts here with the point that, when we
  say something looks red, we seem to be using red
  in the familiar way, as a property of some
  external objects.

19
Logical priority

• For Sellars, being red is logically prior to
  looking red.
• So we cant analyze is red in terms of
  looks-red.
• But X is red iff X would look red to normal
  perceivers under normal conditions is
  necessarily true. If this isnt an analysis of
  is red in terms of looks-red, what is it?
• For Sellars, looks is not a relation (or, its
  a relation if you like to say so, but we
  shouldnt assumed the inferences that would
  follow with other relation words go through for
  looks).

20
The tie shop

• Another vivid example here, of how someone who
  has ordinary colour concepts can come to learn to
  use looks talk.
• John learns to say it looks green rather than
  it is green (and it looks to be blue rather
  than it is blue).
• But what hes learned to report is not a new kind
  of fact about the world.
• What does it looks green report?

21
Experiences and propositions

• When I report, I see that that is green, I dont
  just report an experience I commit to (endorse)
  a claim closely related to the experience.
• This is to apply the concept of truth to the
  experience.
• When I say, instead, I seem to see that that is
  green, I report the experience but I withhold the
  commitment to truth.
• One experience can be a seeing that x is green
  an intrisically similar experience can be merely
  a seeing that x looks green.
• See that and looks talk both make the issue
  of endorsement explicit.

22
What looks come to

• To say that x looks green to s is to say that x
  is having an experience which, if endorsed, we
  would report by saying s sees x to be green.
• But this makes the difference between normal
  reports and looks talk a matter of backing
  away from commitments that would normally be
  associated with the experience being had.
• And it puts in doubt the view that looks talk
  is a way of describing the experience itself.

23
Explaining the link

• Now we can explain the connection between looks
  talk and is talk.
• X is ? iff X would look ? to standard observers
  in standard conditions is necessary because
  standard conditions are conditions under which
  things look as they are to standard observers.
• These conditions are only vaguely specified, but
  thats the way it is with ordinary talk
• This undermines logical atomism, the idea that
  basic concepts are all fundamentally independent
  of each other. On this view we need a lot of
  other concepts (pertaining to standard
  conditions, in particular) before we can have
  (e.g.) colour concepts.

24
Conditions for knowledge

• We dont have even these very basic empirical
  concepts, for Sellars, unless we can not only
  report accurately (this is green, that is red,
  etc.) but are also aware of, and able to detect
  and report, the conditions under which such
  spontaneous reports are reliable.
• Logical atomists can move to sense data reports
  to defend their position. But their commitment
  to sense contents is not supported by anything in
  Sellars treatment of looks talk, so these folk
  will have to haul their own water if they want to
  defend this position.

25
Looking red

• Can we go from somethings looking red to there
  being something red that were aware of?
• First challenge Univocity. The red things look
  to have is the very red some things do have so
  to say that its a sense content that is red
  when something looks red to us is peculiar
  sense contents are very different from physical
  objects, but this line forces on them all the
  sensible properties we ordinarily attribute to
  physical things.

26
Explanations

• If we want to explain why x looks red to s, we
  can often do it by saying, well, s is in C
  circumstances, and x is orange, and in C
  circumstances orange things look red.
• But we can give other sorts of explanation, which
  bring us closer to the heart of sense content and
  sense data talk.
• Such an explanation would focus on sensations
  (the inner postulated entities Sellars has
  already accepted). But this makes the most
  non-theoretical items (for sense-data theorists)
  into theoretical items, driving the wedge too
  early.

27
A more sympathetic account

• Experiences can have shared propositional
  contents that receive varying degrees of
  endorsement (full all the way to none at all).
• The shared content of all these is the
  descriptive content of this account of an
  experience.
• But looks talk only specifies this description
  of our experience indirectly (as in, if conds
  were standard, it would be a seeing that x is
  red).
• So we have an open question What is the shared
  intrinsic character of these experiences?

28
Invoking red talk

• The obvious trick here is to invoke talk of red
  to describe this content.
• But we need to separate red from physical things
  to do this.
• Here we run into the familiar analogy Sellars has
  already invoked in other papers.
• The red of experiences here is not the red of
  physical things, but it is analogous to it.
• Some try to avoid this by separating things from
  their surfaces, and invoking wild surfaces in
  cases of mere looking.
• But we dont really think things have surfaces
  as this two-dimensional, detachable account
  proposes.

29
Impressions and Ideas Logical Concerns

• Ings and eds An experience is often thought
  of as a process that occurs in us this is an
  experiencing, the having of an experience.
• But experiences are also thought of in terms of
  what is experienced in them (the object of the
  experience).
• Sellars is concerned here that we not move too
  quickly from a description of an experiencing to
  conclusions about what it is that is experienced
  in that experiencing.
• In particular, what he calls the common (
  shared) descriptive component of our
  descriptions of experiences may well not be what
  is experienced for instance, if a red item is
  part of this CDC, we may not in fact have
  experienced anything red.

30
Names, labels etc.

• So the question arises Do we really have a name
  for such experiences?
• What about of here? Cant we describe these
  experiencings as experiences of red? Cant we
  add sensation talk here, and talk about
  sensations of red?
• An objection if red is always properly read
  as a property of physical objects, an experience
  or sensation of red demands something red just as
  much as a red experience would.
• But the of we use in talk of ideas, the
  intentional of, doesnt imply the existence of
  the thing an idea is of (nor does believes in)
• This logical feature shared by sensation of and
  believes in does not require further parallels
  between sensation talk and intentional contexts.
• And this is despite the fact that historically
  the assimilation of sensations to ideas was
  widely endorsed.

31
Objective existence

• So for Descartes and many other early moderns,
  just as red was said to have objective
  existence in the idea of a red triangle, it was
  also said to have objective existence in the
  sensation (or impression, in Humes terminology)
  of a red triangle.
• Sensations were simply thought of as more
  detailed and specific than the abstract
  thoughts we form.
• But this is clearly a mistake, and the
  non-extensionality that makes it tempting does
  not give us any real argument for it.

32
Still looking

• So were still looking for an intrinsic
  description/label for these experiences.
• But historically many have taken the view that
  this is just wrong-headed we know exactly what
  kind of experience this is, through our
  privileged inner awareness of it.
• This leads to the problem of other minds, of
  course.
• But its also a clear instance of the given
  this time, what is said to be given is the
  kinds to which our sensations belong.

33
Two questions

• How do we become aware of the sorts that our
  experiences fall under?
• How do we communicate/share the categories each
  of us has for their experiences?
• The view Sellars is criticizing here thinks the
  given is an answer to the first, and accepts
  that there is no answer to the second.
• This answer to the first assumes that our
  awareness of the sorts under which our
  experiences are to be classified is given, a
  feature of immediate experience.
• Recall here Sellars earlier invocation of the
  empiricist notion that all categorization of
  items under sorts requires learning.

34
Universals, repeatables and determinables

• For Sellars a universal is a repeatable (this is
  the std. view today).
• Thus even a very specific/determinate sort of
  experience (my current visual fields character,
  for instance) can be a universal, since this
  arrangement of colours etc. can be repeated in a
  later experience.
• But the tradition (Locke et al.) thought of
  universals as generic, i.e. as abstracting from
  these details, and passed over the problem of
  repeatables to focus on the formation of such
  more general ideas. (Note the nice discussion of
  the need for both conjunctive and disjunctive
  ideas to cope with the phenomena here.)

35
The key point again

• LBH all agree that merely by having sensations we
  have the idea of determinate repeatables or
  sorts.
• A nominalist alternative suppose we think of
  the awareness of a red sensation as the awareness
  of a red particular, rather than as the awareness
  of its being red.
• Then the formation of ideas of repeatables would
  proceed by the association of words (red) with
  a range of resembling particulars.
• But the association needs to be handled
  carefully If we think of it as grounded on a
  prior awareness of the particulars as resembling
  or even as red then the myth of the given rises
  again.
• If we instead reject these, and simply say that
  it is the group of particulars that is associated
  with the word, then we get a kind of nominalism
  it is only from the later point of view, where we
  have a language in which to pick out classes of
  resembling particulars, that we are aware of
  their resemblance before that, all we have are a
  range of patterns of response that we can be
  trained in.

36
The upshot

• This is not Sellars nominalism its pretty
  crude, really.
• But it does break the link between sensations and
  thought-contents, freeing up the direct
  perception account according to which words are
  directly linked to the external world and its
  features, not to our sensations or impressions.
• The essential causal role of red sensations is
  not undermined by this but it is not an
  inferential role in which our awareness of the
  immediately available redness of some
  sensations provides the facts and ideas out of
  which we somehow arrive at the idea of redness as
  a feature of physical things.

37
The logic of means

• The object here is to provide more material for
  Sellars psychological nominalism, the view that
  we have no awareness of logical space prior
  to/independent of learning a language.
• But Sellars rejects the crudely associative view
  of how words are linked to the things they apply
  to.
• On this view, red means what it does because it
  has the syntax (grammar) of a predicate and is
  a response to red things.
• This isnt nearly enough (its too atomistic) for
  Sellars.
• But it wouldnt even be tempting if it werent
  for the naïve word-world view of means contexts.

38
Recalling BBK

• Here we find points familiar from BBK means
  for Sellars invokes translation (and a rich
  normative parallel between the two items joined
  by means as in Rot (in German) means red.)
• This covers the ground in a way that the more
  common view cannot, since it equates the force of
  means in
• Und means and.
• Rot means red.

39
Simple association is not enough

• So the mere invocation of means contexts does
  not generate a theory of how a word comes to play
  a role sufficiently similar to that of another
  for us to say that they have the same meaning.
• In fact, this standard is rich and flexible
  translation is a complex art, and its demands
  vary with our communicative aims/concerns
• In particular, understanding red requires quite
  a bit more than the grammar of a predicate and an
  association with red things. Linguistic
  thermometers are not enough to capture meanings.

40
Foundations

• A foundation for empirical knowledge, for
  Sellars, is a pretty specific thing.
• The myth requires that the foundation be facts
  that are
• Non-inferentially known.
• Such that knowledge of them presupposes no other
  knowledge.
• Such that all knowledge of empirical matters of
  fact is ultimately decided by reference to
  knowledge of these foundational facts.

41
Something emphasized

• Some argue that, if a piece of knowledge depends
  on having some other knowledge, it must in fact
  depend on inference in some way.
• For Sellars this is a mistake, and its part of
  the myth. The kind of dependence that his holism
  invokes does not make all our knowledge a product
  of inference.
• But it does make its content dependent on our
  grasping its inferential relations to other
  claims.

42
Details

• Observation reports seem to have authority,
  without being inferred from other claims we
  already believe.
• How shall we understand this authority?
• One key is the link between reports and
  circumstances.
• This point turns on the distinction between
  fact-stating uses of sentences and report-making
  uses.

43
Token-reflexives

• Words like here, now, this or that are
  token-reflexive they link up to their
  circumstances of utterance in ways that lead the
  same sentence to make different assertions in
  different circumstances.
• Tensed verbs have much the same effect, since
  they are tied to times of utterance too.
• Reports regularly, though not necessarily, make
  use of these sorts of linguistic device to link
  the report made to the circumstances it is a
  report about.
• A report must have some kind of direct link to
  the circumstances it reports about
• Fact-stating doesnt depend on circumstances in
  the same way.

44
Two sources of credibility

• Some sentence tokens are credible because they
  are tokens of a type all tokens of which have
  authority.
• But some sentence tokens are credible because of
  how they are linked to the circumstances in which
  a token is produced The token came to be
  through a certain process under certain
  circumstances.

45
Authority for sentence types

• Sellars says some sentence types are
  intrinsically authoritative some analytic
  sentences belong to this group.
• Some get authority by inferential links to
  others.
• But empirical knowledge will not reduce (without
  some residue) to these two groups. (No empirical
  sentence type has intrinsic authority, and
  inferential links to such wont do the trick
  either.)
• So we need sentence types whose authority comes
  from tokens that are produced in the right way
  and in the right circumstances.

46
The flow of credibility

• These two sources of credibility flow in opposite
  directions
• Credibility for analytic sentences flows from
  types to their tokens, and then via inference to
  some other sentence types.
• Credibility for observation sentences flows from
  tokens to types, and then again via inference to
  other sentence types.

47
Sellars on Empiricist Dogmas

• Conflating the authority of analytic and
  observational sentences An attempt at a unified
  theory of credibility leads to trouble.
• The idea is to appeal to rules just as the rules
  for using the concepts in an analytic sentence
  ensure that it must be true, the rules for using
  the language involved in an observation report
  are such that, if they are followed when the
  report is made, the observation report is true.

48
Prefatory points

• Reports can be thought of as stripped of their
  action and interpersonal aspectsa mere,
  spontaneous thinking that p can be a report.
• But thinking of them as actions has influenced
  the tradition here, which construes the rightness
  of a report as a special case of the rightness of
  actions (under some system of rules).
• When we think of them this way, we have to
  construe them as deliberate followings of these
  rules (otherwise the rule is a mere
  regularity).
• That is, we have to judge that the circumstances
  were such that the rule requires the report and
  make the report for that reason.

49
The given again

• This leads directly to the given again, since
  these observation-actions are correct only when
  they are made in the course of following a rule,
  which requires that we be able to judge that the
  circumstances are such that the rule
  requires/endorses the report.
• Such judgments depend on a primitive awareness of
  the matters of fact necessary to apply the rules
  for using these words.
• Recall that we are talking here about the
  foundational judgments that are at the root of
  all empirical knowledge so there is no escape...

50
But is there a way out?

• Begin with mere reliable reporting the
  thermometer account.
• Now, can we extend this to cover something more
  properly like knowledge?
• The key question is authoritybut authority for
  what?
• Actions are not what we need here mere behaviour
  of a type that its reasonable to endorse is
  enough (action presupposes too much
  deliberateness here).

51
Being a knower

• The key, however, is the subjects own ability to
  recognize that she is in a position of authority
  with respect to such reports.
• That is, she must be in a position to recognize
  that she is a standard observer in standard
  conditions, so that her inclination to report
  this is green is indeed reliable.
• This requires quite sophisticated conceptual
  grasp of the place of such reports in the world
• It also seems to require a kind of induction.

52
The threat of a regress

• The idea is that to make a report that expresses
  observational knowledge, one would have to know
  that such overt verbal performances are reliable
  indicators of the facts reported.
• And this means that all observational knowledge
  depends on a fair bit of knowledge that. In
  fact, we need to know things like X is a
  reliable symptom of Y.
• So observations dont stand on their own feet
  individually.
• Worse, it seems to run counter to the empiricist
  notion that the sort of general knowledge we have
  when we know X is a reliable symptom of Y depends
  on having a lot of knowledge of particular cases
  where X is correlated with Y.

53
Responding to the regress

• This seems to be a regress- before we can
  actually know, by observation, that we have a
  case of Y, we need to know that X is a reliable
  symptom of Y.
• But we need to know cases of Y (and cases of X)
  to justify the claim that X is a reliable
  symptom of Y. OUCH!
• But when we classify an episode or event as a
  case of knowing, we are locating it in a
  normative space. Its possible to establish the
  normative status of an episode retrospectively.
  The regress dissolves (p. 169).

54
Rejecting the foundation

• Sellars rejects the idea of observational
  responses that are both direct and immediate.
• Such observations would be self-authenticating,
  that is, they would stand on their own but they
  are, at root, non-verbal, internal episodes.
• And this authority is transmitted to linguistic
  acts (reports) because of an awareness of the
  rules governing how these episodes are to be
  characterized in words.
• Once we have the force of these episodes cast in
  linguistic terms, inference connects them to
  other sentences, so that these episodes become
  the foundation for all our empirical knowledge.
  

55
Launching the constructive phase

• The point of the rest of the paper is to explain
  Sellars own views on the nature and role of
  inner episodes.
• The sense in which they are non-verbal does not
  (says Sellars) lend any support to the idea of
  the given.
• But Sellars rejection of this particular view of
  the foundation of empirical knowledge is not a
  rejection of the broader idea that there are
  observation-reporting sentences that serve as the
  touchstone for empirical knowledge. It is aimed
  at the idea of the given (which is related to the
  demand that the foundation be locked in and
  independent of any other knowledge).
• The active element of self-correction is crucial
  here.

56
Philosophy of science

• Sellars is concerned about the effects of
  specialization in particular, the recognition of
  philosophy of science as a distinct
  specialization in philosophy.
• The worry seems to be that specialization may
  lead other philosophers to leave science to the
  philosophers of science, and lose track of the
  need for all philosophers to respond to the
  development of the scientific world view.
• Science is the flowering of an aspect of
  discourse that is part of the ordinary world
  view, and which may be lost sight of if we dont
  attend to science and how it works in our
  philosophical thinking.

57
Evaluation

• To evaluate features of our conceptual frameworks
  requires an appreciation of science as revealing
  standards of evaluation and alternative
  frameworks that may well do better in terms of
  those standards.
• This needs care the tendency to say things like
  physical objects are not really coloured is a
  confused way of recognizing that a framework that
  includes such objects has deficiencies that are
  put right in a different framework, now under
  construction in the scientific enterprise.
• On these questions, science is the measure.

58
The given as a barrier here

• If we take some aspect of the manifest image as
  given then this kind of critique is blocked a
  framework anchored in the given is one were
  stuck with (we can alter it in various ways, but
  the basics are fixed).
• This anchor is sometimes thought of as an
  ostensive tie linking the concepts of the
  framework to a aspect of reality. But consider
  the oddity of the idea of an inner ostension.
• This goes with the idea that science is not a
  complete framework in its own right, but a mere
  auxiliary.

59
As if

• The suggestion is that the unquestionable nature
  of the given locks us into any framework that
  it arises in.
• The upshot is that an alternative framework which
  outperforms the given framework must be treated
  in terms of as if, that is, the things of the
  given framework can be thought of as behaving as
  if some other framework were true but we can
  never dispense with the given framework in favour
  of this new (and superior) one.

60
The other error

• The other barrier to what Sellars sees to a
  correct appreciation of the scientific framework
  is the reification of a methodological
  distinction between theoretical and
  non-theoretical discourse into a substantive
  distinction.
• The contrast
• Science as a peninsular offshoot of ordinary
  discourse vs.
• Science as an expression of a fundamental aspect
  of discourse which is central and ultimately
  transformative.

61
Private episodes

• We still have no take on how it is that the
  experience that is a seeing that x over there is
  green is similar to one that is merely a seeing
  that x over there looks green.
• Two routes
• Introduce inner episodes as theoretical posits
  and develop an account of the similarity within
  that theory.
• Posit these similarities as given.
• The second is now ruled out, so we need to turn
  to the first, however peculiar it may seem.

62
Applying the critique of the given

• red etc. apply to physical objects first, so
  while we tend to describe these experiences using
  such words, this use of those words cannot be
  univocal.
• If we just use these words to link the
  experiences to standard occasions on which such
  experiences occur, then we have a definite
  description of such experiences that picks them
  out, but nothing to say about what they are
  really like.
• To be able to notice a certain kind of thing (as
  being of that kind) is already to have a concept
  of that kind of thing. (176)
• One central puzzle The combination of privacy
  (only I can report my inner states) with
  intersubjectivity (others can know about them in
  other ways/ inferentially). Sellars says we can
  have both!
• But before dealing with sensations, Sellars turns
  to consider thoughts.

63
Thoughts

• Even if we take the view that thoughts are
  linguistic, the number of cases where overt
  linguistic behaviour accompanies behaviour that
  it seems to explain is too few the sorts of
  behaviour that we explain in terms of thoughts
  are often unaccompanied by any overt linguistic
  behaviour.
• And when that behaviour is not habitual (more or
  less automatic), we can hardly resist invoking
  thoughts to explain it.
• The classical account holds that there are inner
  episodes called thoughts, and linguistic
  behaviours are meaningful insofar as they express
  those thoughts. Further, these episodes are
  introspectible and even transparent (self-
  revealing/ known whenever they are had).

64
The core of the notion

• We each have a stream of these inner episodes.
• We have privileged, but not infallible or
  complete, access to them.
• They occur, sometimes, without overt verbal
  behaviour.
• We dont need a performance to be going on that
  we perceive inwardly here.
• Thoughts are a sort of linguistic episode, that
  is, they are modeled on over speech.

65
The Ryleans

• Here we imagine our ancestors, speaking a
  somewhat limited language.
• The language describes external objects in terms
  of their sensible features.
• Sellars also supposes that it has semantic
  concepts so they can speak of meanings (read
  in terms of the roles of words phrases)
• The aim is to explore how a language like this
  can come to have the means to talk of inner
  episodes.

66
On semantics again

• The force of semantic talk includes implications
  about the typical causes and effects of certain
  utterances.
• But it is not exhausted by these implications.
• More importantly here, having semantical talk
  allows us to say a lot of things about utterances
  that are also commonly said about thoughts (that
  they are about, or refer to, something, etc.)
• We could try, then, to capture thoughts just by
  appeal to a conditional/hypothetical link to
  overt language and its semantics.

67
The Alternative?

• Inner episodes (rather than conditional/hypothetic
  al speech).
• The classical notion also makes these episodes
  the primary vehicles of meaning but Sellars will
  reject this aspect of the classical view.
• So the challenge is to combine a speech first
  view of meaning with an account of thoughts as
  real inner episodes.

68
Theories and Models

• Here we encounter the philosophy of science
  again, which Sellars will draw on to explain his
  view of the status of these inner episodes.
• We typically develop a theory by positing certain
  entities and certain postulates about how they
  behave.
• These are then linked to some observable
  phenomena (regularities or empirical laws) in
  effect, we propose that the phenomena are as they
  are because they involve these items in a certain
  way.
• But in fact theories are typically developed in a
  very different way.
• We generally draw on known behaviours of familiar
  objects (waves, for example), and present these
  as a model of what is going on with respect to
  some phenomena.
• Such models are qualified (we dont propose that
  electromagnetic waves are just like waves in
  water, but that they are similar in some
  respects, while different in others.)

69
Continuity

• Further, common-sense is not limited to
  induction. (Unlike the idealized MI, it can
  invoke postulated entities too and for Sellars,
  the MI does contain fossilized posits, despite
  its limitation to inductive methods.)
• So our Ryleans will have the ability to develop
  postulational explanations for similarities and
  differences in the behaviour of physical things.

70
Behaviourism

• Note that this was a major influence in
  psychology when Sellars was writing EPM.
• Sellars rejects narrow forms of behaviourism,
  while being perfectly happy with methodological
  behaviourism.
• This does not require that psychological concepts
  be analyzable into behavioural terms, but only
  that assertions in our new, scientific
  psychological language be justified by
  behavioural evidence.
• So, while we introduce these concepts by means of
  their links to behaviour, we do not suppose that
  their meaning is exhausted by this link to
  behaviour they can be fully fledged theoretical
  concepts.

71
Expectations

• Of course we expect any such theory to link up to
  our physiological understanding of the nervous
  system and its role in behaviour.
• But this does not mean that the behaviourists
  concepts are physiological at the outset
• It does mean that our concepts are both linked to
  behaviour and linked to our expectations of how
  they will fit in with other aspects of our
  understanding of the world.

72
Back to Jones

• Jones proposes a theory of inner episodes,
  modeled on overt speech.
• These explain intelligent behaviour that goes on
  even when no overt commentary accompanies it.
• The episodes carry the semantic force of the
  speech acts they are modeled on.
• And they are taken as the real, proximate
  causes of intelligent behaviour, even when
  accompanied by an overt explanatory discourse.
• Think here of how we accompany demonstration with
  commentary when instructing people in how to do
  something

73
More details

• The commentary follows, explaining the parallels
  between speech and thought and where the two
  differ.
• One important point Nothing in the theory
  requires that we have thoughts first and language
  second the ability to have thoughts may depend
  on learning a language.
• Another The theory does not imply that thoughts
  are the truly semantic items, with utterances
  only semantic in a derivative sense, as
  expressing certain thoughts.
• Finally, these inner episodes need not be
  construed as immediate experiences no perceptual
  model of them is implied here.

74
Introspection

• Having been taught to use this theory, people
  find that they can sometimes report their
  thoughts without relying on the overt behavioural
  evidence others need to reach the same
  conclusion.
• Now the theoretical language has a reporting
  role, and its one that recognizes a kind of
  privileged access.
• But the concepts involved are still public, and
  their link to behaviour is still part of their
  meaning so the privacy involved here is not
  absolute or incorrigible.

75
Inner perceptual episodes

• Now were ready for Jones second theory inner
  episodes modeled on the perceptible features of
  objects, which are invoked to explain illusions
  and related perceptual phenomena.
• This provides us with a theory in which we really
  do have a description of the inner episode that
  is the same when we see a red triangle, see a
  triangle that looks red, see something that looks
  like a red triangle and merely seem to see a red
  triangle.
• The price, of course, is to accept that these
  inner states are postulated, not given.

76
Details

• These are episodes (goings-on) not particulars.
• They should not be construed in
  perceptual/epistemological terms The model is
  not, a seeing of a red triangle but a red
  triangular surface.
• In general, we have the occurrence of inner
  replicas a matter of a description of whats
  going on in us, not normative matter at all.
• The terms the theory provides give us a
  substantive description of these states, rather
  than the mere definite descriptions used so far.

77
The content of the description

• The content of the description is a matter of the
  theoretical language and the inferences it
  supports.
• Thus we know that the relations of compatibility,
  incompatibility, etc. grounded in the ways
  objects sensible properties can be arranged are
  shared by these theoretical predicates describing
  our impressions/sensations.
• But the limits of the parallel are not clearly
  specified yet, and can be worked out/explored in
  a pretty open-ended way.

78
Contentless form?

• Sellars objects to the view that such concepts
  (characterized by their internal inferential
  relations and by a link to certain
  observational circumstances) are themselves
  contentless.
• For him they are no more contentless than any
  other theoretical predicate,
• And the view that all such predicates are formal/
  contentless is a part of the myth of the given
  (according to which only the predicates that are
  part of the given have real (ostensive) content.

79
Expectations

• What is the relation between these states of
  perceivers and physiological models of ourselves?
  
• In the long run we expect them to fit together,
  giving a physiological account of these states.
• But our present physiology does not have the
  resources to do this (homogeneity again).
• And Sellars doesnt believe that the physics of
  non-living things has the resources either
  something unique to organisms (ones sufficiently
  like us) needs to turn upwe need a micro theory
  of sentient organisms, he says, to find the
  micro-states corresponding to these posited molar
  states.

80
The tradition and the sequel

• The modern view that objects arent really
  coloured, and colour is in the mind of the
  perceiver, is a confusion of this scientific
  anticipation with a feature of the manifest
  image.
• The upshot of Jones work is that we have a
  notion of both thoughts and impressions that
  allows introspection (individuals are the
  reporters for their own states) but also makes
  the link with observable behaviour (and thus
  training by our linguistic seniors as we grow up)
  part of the meaning of this language.

81
The order of things

• The language of impressions is no more the result
  of antecedent noticings of impressions than the
  language of molecules is the result of antecedent
  noticings of molecules.
• The phenomenologist, in describing these states,
  fails to recognize her own creative role
• She takes herself to be analyzing human knowledge
  as it already was, when she is actually
  contributing to an entirely new body of knowledge
  whose projected links to a scientific
  micro-theory of sentient organisms point towards
  a unified science including impressions.