Phenomenalism

1
Phenomenalism

• Stopping the pendulum?

2
The general idea

• Classical Idealism (Berkeley) Physical objects
  are actually (structured) collections of ideas.
• More recent (and more modest) efforts Carnap,
  Die Logische Aufbau der Welt. Carnap tries to
  come up with an analysis of talk about the world
  in terms of a. Logical structure b. Space and
  time c. Sensible properties. (Red at place p
  and time t as a protocol sentence.)
• Others Attempts to analyze ordinary object talk
  as some kind of theory grounded in immediately
  observable features of our sensory experiences
  (some think there really is a parallel here with
  scientific theories).

3
Direct (Naïve) Realism

• while the verb to see has many usesits
  primary use is one in which a person is said to
  see a physical object and to see that it is of a
  certain colour
• Sellars winds up defending some such view (but
  thats later).
• A more naïve realism holds that things seen
  (whatever they may be) always are as they appear
  to be this Sellars does not hold, so he adopts
  direct realism to distinguish the view he will
  defend from this more naïve realism.

4
The phenomenalistic theme

• Objects really do have the sorts of properties
  they look to have.
• Direct realism holds that the standard use of
  colour predicates is in contexts like O is red
  at place P.
• Looks-talk must then be explained in terms of
  this basic sort of claim.
• Classical phenomenalism has a different starting
  point A collection of more basic objects
  which have these properties in a more basic
  way. Ayer calls these objects sense contents.
• The idea is that a physical object is redp, i.e.
  red in the physical sense, if and only if there
  are reds sense contents that are parts of it
  under normal viewing conditions.
• The object merely looks red when there are reds
  sense contents that are parts of the object, but
  are not produced under normal viewing conditions.
  

5
Saving sensory judgments in cases of illusion

• The main point for the phenomenalist is that when
  something merely looks redp, this is because
  there really is a reds sense content being
  experienced.
• Further, all the sense contents produced by the
  object, under normal and abnormal viewing
  conditions, are parts of the object.
• The DR, on the other hand, says that only parts
  of physical objects have colours and physical
  objects and their parts are publically
  observable, i.e. they belong to the
  intersubjective world.

6
Surfaces and slippery slopes

• It seems right to say that, when we call (say) a
  car red, what we mean is that it has a red
  surface.
• Such a surface is not 2-dimensional its a 3D
  physical object, the (thin) layer of paint that
  covers the car.
• But its tempting to move in a more metaphysical
  direction, and say that what really has a colour
  is a really 2D surface, and to conclude that what
  we really see is such 2D surfaces (the rest of
  3D physical objects being inferred from the
  surfaces we see).
• Even if we make this move, were still operating
  in the public world were not yet
  phenomenalistseven though we may now describe
  physical objects as made up of actual and
  potential (exposable) colour surfaces.

7
Rejecting surfaces

• There are a few ways to relate this talk of
  surfaces back to the colours of 3D objects.
• But its not pretty we can make lots of
  different exhaustive slicings across an apple,
  and to say the apple is made of all of these
  seems redundant at best.
• It seems we should either say the 3D object comes
  first, and its slices are dependent and only
  actual when the thing is sliced, or the object is
  really a collection (continuum) of colour points,
  and both the object and the slices are (3D and
  2D) collections of such points.
• The last seems dubious the first makes the
  slices depend on the 3D object.
• The real mistake Sellars is worried about,
  though, is the distinction between what is really
  seen, viz. the surface and what we then
  believe/infer from it (the object). This is an
  epistemic step towards phenomenalism, since it
  leads to the idea of unattached surfaces that
  we can also see, and to an epistemic conservatism
  about what we see that makes the phenomenalist
  account of illusions tempting.

8
Still some distance to go

• Pure seeing is introduced here for the new D.R.
  view that says, what we see (really, purely) is
  just surfaces, and we then infer the objects that
  have them.
• But so far, the D.R. says there is no dagger that
  Macbeth sees, and this new D.R. agrees.
• However, the new D.R. may be tempted to say there
  was a dagger-shaped surface there (with no dagger
  inside it?).
• This step removes the surfaces from the public
  world and shifts them to the theatre of the
  mind where the classical phenomenalist is at
  home

9
Resistance

• The new D.R. need not go this way she can keep
  her new surfaces firmly linked to the public
  world.
• On this view, surfaces without cores just dont
  exist. And (consequently) neither do surfaces
  without backs.
• On the other hand, if she begins to say that only
  facing surfaces really exist, the new D.R. is on
  her way to phenomenalism.
• For Sellars this is just a mistake we do not see
  surfaces and infer objects we see objects,
  though what we see of them is just a facing
  surface, we take it that there is an object and
  that it has a back (and inner surfaces, if we
  should slice it) with some shape and colour too.
• Both the objects and the parts of them that we
  see can seem to be other than they are Illusion
  remains possible concerning them.
• But a second train of thought leads to
  phenomenalism Here, what is basically seen is a
  sense content these are private, and not subject
  to illusion they are always as they seem to be
  to the subject.

10
Retaining a key point

• Rejecting phenomenalism does not require that we
  reject the existence of phenomenal colour
  expanses as elements of visual experience.
• Sellars is persuaded that there are such things,
  but that we need to locate them very carefully in
  our conceptual scheme.

11
Phenomenalism The slogan

• Physical objects are patterns of actual and
  possible sense contents.
• Taxonomy Three traditions on the nature of
  sense contents.
• The first Grounds its talk of sense contents in
  ordinary perceptual talk directly sees and
  directly sees that parallel normal use of
  sees and sees that. So direct seeing
  involves knowing things about what is directly
  seen.
• For example, the usual inference from S saw an
  X to an X exists carries over S directly saw
  a red triangular sense content implies there was
  such a sense content.
• Further, just as objects can exist unsensed, this
  model allows (at least in principle) that sense
  contents can also exist unsensed, and just as
  objects can look other than they are, so can
  sense contents (error is not ruled out). (Contra
  Berkeley here.)
• But of course direct seeings are seeings of sense
  contents, not of public objects.

12
A second approach to phenomenalism

• Links sense contents to conceptual thinking.
• This is where we get the esse is percipi notion
  Just as, for there to be an idea of x, someone
  must be thinking of x, for there to be a red
  sense content, someone must be having
  (perceiving) it.
• That some red triangular expanses (the
  sense-content ones) must be perceived to exist
  does not mean that all are like this we can
  still have them out there in the physical world
  too.
• So its a further step to say (with Berkeley)
  that the real red triangular expanses are all
  dependent on perception, and that there are no
  independent ones at all.

13
Tangling the first two

• Someone who starts in the first way might still
  move on to claim that for sense contents esse is
  percipi.
• But the argument she would need to make her case
  must either be inductive (i.e. all the red
  triangles observed have been perceived, so in
  general all existing red triangles must be
  perceived?) or synthetic a priori (something
  about the idea of a red triangle demands a
  subject who perceives it this sounds rather like
  Berkeley).
• On the second approach, though, X is red
  outside of contexts of the form X is a red sense
  content is just ill-formed. This emphasizes how
  far from our ordinary perceptual talk the second
  sort of theory takes us.
• Further still X sees that has no parallel on
  the second approach there is no X senses that
  here, just direct-object sensings of various
  kinds of sense contents.

14
The third phenomenalism

• Here the beginning point is the link between
  sense content talk and appears talk.
• So S senses a red triangle doesnt imply that
  there is a red triangle (as with approach 2).
  (After all, that there appears to be a red
  triangle does not imply that there is one!)
• We can get a form of the implication back by
  force.
• According to the third version, though, is that
  what we directly know in sense perception is
  facts about sense contents, i.e. (on this
  account) facts about how things appear to us.
• But this makes classical phenomenalism hard to
  accept, since it puts appears talk in its
  application to the physical world first, and
  sense content talk is defined/understood in
  terms of a language that apparently presupposes
  the categories of public physical objects and
  their sensible properties.

15
A thesis

• Whenever there appears to S to be a red
  triangular physical object somewhere, then it is
  also true that S has a sensation of a red
  triangle.
• 3 gives us this by defining the second in terms
  of the first. But a phenomenalist must find a
  different route to this claim.
• Sellars also endorses the claim, but not in a way
  that gives comfort to the phenomenalist.
• Sellars will defend a form of the second line on
  sense contents.

16
A forced choice

• The phenomenalist has to choose between a version
  of the second account of sensations (one that
  doesnt equate sensations with states produced
  ordinarily by talk, which makes them dependent
  on ordinary object talk), and the first (in which
  sensations are perceived directly, and such
  perceivings involve direct knowledge of our
  sensations).

17
Esse is Percipi

• On the first approach, esse is percipi is not a
  natural principle to adopt (since the model is
  ordinary perceptual talk, and there what is
  perceived can and does exist without being
  perceived).
• It can be added by force, if a direct object
  version of sensing (S senses x) is added, and we
  assert that sensations only exist in such
  relations to subjects. But we lose a direct link
  to the cognitive side here such relations might
  exist, in principle, without S knowing anything
  about the sensation S is having.

18
Towards a refutation

• The challenge to phenomenalism comes here.
• Recall that phenomenalism holds that ordinary
  objects are (in fact) collections of actual and
  possible sense contents.
• So we need to understand what possible sense
  contents are.
• The suggestion Sellars makes is that they are
  possibilities that subjects are in a position to
  bring about analogous to skids that a driver is
  in a position to bring about.
• So, for instance, when I have my eyes closed and
  Im standing in front of a white wall and its
  normally lit and I can open my eyes, then theres
  a possible white sensation that I would have if I
  opened my eyes.

19
Justifying such claims

• To justify that sort of claim about what
  sensations I would have if I were to do
  something, I need induction.
• This requires that I notice a regular pattern in
  my sensations, of the form Whenever
  circumstances are C and I do A, E results.
• We can think of many instances of such
  generalizations where E describes sensations that
  I will have.
• But the standard cases are cases where C and A
  are specified, not in terms of the sensations
  that I have or had, but in terms of ordinary
  objects and actions/movements that I make in
  public space (for example, the fireplace on p. 79
  or the white wall above).

20
A there pure phenomenal generalizations?

• The circumstances and the action must be
  described in phenomenal terms, not just the
  resulting sensations.
• But will the circumstances and the action be
  described using actual sensations, or actual and
  possible (i.e. conditional) sensations?
• On one hand, the phenomenalist has only claimed
  to reduce objects goings-on in the physical
  world to a combination of actual and possible
  sensations.
• But on the other hand, if all the generalizations
  here invoke both actual and possible sensations,
  how could we ever have learned them? That is,
  what we learn about the world depends on actual
  experience possible experience (whether
  phenomenal or not) makes no impression on us and
  teaches us nothing.

21
Getting down to actual sensations

• So the challenge to the phenomenalist is to frame
  generalizations that are pure, in the sense that
  they are stated solely in terms of actual
  sensations only then could we have inductive
  evidence of what sensations are possible in a
  given (purely phenomenally described)
  circumstance.

22
E and A generalizations

• Sellars allows that there really are some purely
  phenomenal generalizations.
• But he distinguished between two kinds of such
  generalizations essentially autobiographical
  and accidentally autobiographical.
• The first are generalizations that we would
  normally explain as due to the fact that, as
  individuals, we live among particular objects and
  are subject to other particular ordinary object
  perceptual facts about ourselves. They cannot be
  separated from our autobiographies.
• The second, however, though they are learned in
  the course of our experience as individuals, hold
  independently of the particular circumstances of
  individuals only these could provide a general
  account of possible/conditional sensations in
  pure terms that would support a phenomenalistic
  reduction of ordinary objects to actual and
  possible sensations.

23
Details on what generalizations are needed

• what the phenomenalist wants are
  generalizationswhich are accidentally
  autobiographical, generalizations in which the
  antecedent (circumstances C/ available/possible
  action A) serves to guarantee not that I am in
  the presence of this individual thingbut rather
  that my circumstances of perception are of a
  certain (general/sensation-only dependent) kind.
  (83)

24
Theres more

• A lot has been granted that could be disputed
  here the idea of persons is also part of the
  ordinary object framework, and it too needs to be
  reconstructed (ditto for their actions the focus
  here has been on the circumstances C).
• E generalizations come with dirty hands (84),
  i.e. they hold only for people in circumstances
  that are fixed in terms of the types of objects
  present (including their sensible properties and
  spatial/temporal arrangements).
• So they are not credible as unrestricted
  inductive generalizations which are what the
  phenomenalist really needs.
• There are real generalizations here- but they
  make the possible sensations available to a
  subject dependent on the physical circumstances
  of the subject. That is, they depend on the
  framework of ordinary objects and our beliefs
  about how these objects and the subjects
  relations to them determine what sensations the
  subject will have.

25
A hypothetico-deductive turn

• What is the hypothetico-deductive (HD) method?
• Theories are not arrived at by direct induction
  from our observations they are hypotheses which
  can subsequently be tested by what they imply
  about our observations.
• So the idea is that we form hypotheses and deduce
  consequences for observations (Then we observe to
  see if those consequences actually hold).
• Sellars argument to here has assumed a kind of
  inductive basis for the bootstrap attempt to get
  ordinary object concepts out of sensations.
• Maybe an HD approach can save phenomenalism!?!

26
Whats involved

• Neo-Lockean Physical objects are part of a
  theory that explains observed facts about our
  sensations.
• No more translation of physical object talk
  into phenomenal talk.
• Nevertheless, theoretical entities (such as
  ordinary objects) are still just that purely
  conceptual posits, intended to do the job of
  organizing our observations. There is no need to
  regard ordinary objects as anything more than a
  convenient hypothesis.

27
Sellars response

• For Sellars, we should take theoretical entities
  seriously, not dismiss them as mere conceptual
  tools.
• But this is not his reason for rejecting
  neo-phenomenalism.
• The relation between observation-language and
  theory that HD requires is that inductively
  confirmed generalizations in the observation
  language correspond, via the bridge rules, to
  theorems in the theoretical language (and no
  theorems should correspond to disconfirmed
  inductive generalizations).
• But this breaks the proposal Sellars earlier
  argument against classical phenomenalism showed
  that there are no such confirmed generalizations
  in the language of sense-contents.

28
Where from here?

• Return to direct realism we directly see real
  (public) objects, and in seeing them, we see that
  they have various sensible properties (that is,
  seeing in this sense is an epistemic/
  knowledge-producing/ judgment-involving act).
• No other form of knowledge is more basic. (This
  is where our knowledge of the world begins.)
• But this does not mean that we dont have direct
  knowledge of other things.
• Including that we seem to see something/ and
  features of our sensations/ visual impressions.

29
Direct?

• Direct knowledge is non-inferential that is, we
  dont acquire it by reasoning from other things
  we know.
• But more knowledge that p requires (on the point
  in question) authorization, a right to be
  convinced that p.
• The inference schema for direct knowledge is
  Xs thought that-p occurred in manner M. So
  (probably) p.
• A label This is trans-level inference, because
  it involves a shift from the meta-level at which
  we talk of thoughts, their contents, and the
  conditions in which they arise, to the level of
  endorsing one of those thoughts.
• Care is needed to distinguish what is and is not
  directly known in familiar cases of perception
  the idea, for Sellars, is to avoid
  representationalism. (?)

30
Directness vs. Security

• There is a temptation to identify directness with
  security, and say that we know more directly when
  there is less chance that were wrong.
• Sellars rejects this directness is a matter of
  whether nor not inference is required. So, if I
  directly see a book and see of it that it is red
  and rectangular on the facing side, that is
  direct knowledge of the book. I might then infer
  that its also red and rectangular on the hidden
  side, but that is not direct knowledge.
• But my knowledge that Im seeing a book is less
  secure than my knowledge that I seem to be seeing
  a book, though it is not any less direct.
• Even if I sometimes do infer facts about what Im
  seeing from facts about how things appear to me,
  this doesnt mean that all knowledge about the
  physical objects I see is inferred in this way
  (in fact, I learn to make claims about the
  sensible features of objects around me directly
  long before I learn to talk about how things
  appear to me.)

31
Parasitism

• In fact, Sellars claims that the frameworks of
  qualitative and existential appearings and of
  sense impressions are parasitical upon discourse
  concerning physical things.(89)
• The phenomenal world of these things and their
  perceptible qualities does obey regularities, and
  provides an observational starting point for
  scientific theorizing.

32
Learning to perceive the world

• An abstractive account of concept-formation holds
  that we come by concepts like red by means of
  experience of red things. (This makes the notion
  that having the sensation of red is the key, and
  even that the sensation of red is the original
  from which the idea of red is abstracted. See
  Hume!)
• Sellars claims instead that coming to have a
  perception of something red as red, that is, a
  perception that involves the judgment this is
  red, is not just a matter of having the relevant
  sensation.
• It requires also a rich background of language
  learning and dispositions to accurately report
  the colours of things and to distinguish one
  colour from other colours, this thing from that
  thing, etc.
• This logical space of things and their properties
  is an evolutionary development, culturally
  inherited. (90)

33
The process

• When I know something about an object by
  perceiving it, there is a link between the object
  and my knowledge of it the sensation it causes
  in me.
• This sensation is what I respond to, but its
  effect is causal I dont infer from facts I
  directly perceive about my sensation to
  conclusions about its cause.
• Instead, my having the sensation is one causal
  factor (another is my possessing the right
  conceptual framework to respond correctly to the
  sensation in those circumstances itself a
  product of normal development and
  language-learning) leading me to judge that some
  claim about the perceptible qualities of the
  object is true.

34
A familiar idea, again

• Sense impressions are postulated, in order to
  explain how things appear to us.
• But for Sellars, if a theory is a good theory, we
  should believe in the entities that theory
  posits.
• So for Sellars, there really are sense
  impressions.
• But they are not sense data, that is, they are
  not cognitive states that form the premises on
  which our theory of everyday things and their
  sensible properties is built.
• Here we should pause to consider the pragmatic
  theory of observation.

35
Learning the language of sensations

• To learn to describe our sensations, both we and
  our instructors must know
• How to report/ describe the sensible properties
  of things in the physical world.
• How we describe sensations by analogical use of
  these sensible properties.
• Then we can be trained to use seems talk and
  other ways of describing our sensations, and even
  to report our sensations directly.

36
Two links between sensations and physical objects

• A sensation of a red triangle is a sensation that
  is normally caused, under standard conditions, by
  the presence of a red triangle before our eyes.
• Impressions of red/blue/yellow triangles resemble
  and differ in ways analogous to the ways
  triangles that are red/blue/yellow resemble and
  differ.
• The same goes for different shapes of a given
  colour.
• So we get families of predicates for properties
  of sensations, each based on the predicates for
  the corresponding sensible properties of physical
  things.
• Since these new predicates are predicates for
  features of an episode (which is occurring in a
  person), i.e. features of something that is going
  on, they are correctly said to be adverbial. They
  pick out kinds of sensings.

37
But one more thing

• A theory aims to account for the inductively
  supported generalizations belonging to some
  observation language.
• Why believe in its entities, if its really just
  used as a calculational device to capture such
  generalizations?
• Sellars says, if the generalizations really hold
  up to the level of epistemic variance, then
  there really is no reason to take the theory as
  more than this.
• But 1. We dont really know the things of the MI
  exist, as we conceive them in the MI.
• And 2. If they dont, then theres no guarantee
  that the laws of a successful theory will
  correspond to inductively justified
  generalizations in the MI.
• Instead, Sellars suggests that the SI will
  explain why the things of the MI obey certain
  generalizations, to the extent that they do. No
  strict generalizations cast in MI terms survive
  detailed careful examination.

38
Towards Scientific Realism

• In principle, the SI could replace the MI in all
  its uses, from observation to decision-making.
• But this is not the time The MI provides a
  constant background against which we can
  check/compare the successes of scientific
  theories. We dont want to lock in a particular
  body of scientific theory yet.
• But we still have some questions here. Just what
  do we do with the (occurrent) properties of
  physical objects, if we dont think anything
  really has them, but we describe our sensations
  in terms of an analogy with them?

39
3 Stages

• 1. Nothing is really coloured (only public
  objects could be, and they are not).
• 2. Sensations will persist, in some form, in the
  SI but weve said they have properties analogous
  to colour (etc.). So colours in some sense may
  persist in the SI as features of states of
  conscious, perceiving organisms.
• 3. Persons in the MI are single logical subjects.
  Thoughts and sense impressions, in particular,
  are attributed to a single subject. The
  framework in which we think this way must be
  reconciled with a scientific description of
  persons as complex structures built of many
  separate parts.

40
The place of persons

• But because these complex structures think of
  themselves as subjects, the single logical
  subject, a person, is important in their thinking
  and behaviour. (The neo-Hobbesian position comes
  in here 101f)
• This is no longer a matter of describing what a
  person is, but instead a matter of identifying a
  subject for normative purposes.
• To cope with sensations now, in our descriptive
  theory, we need to separate them from the
  grammatical subject (the I) who is said to have
  them.

41
Counterparts

• When we think of two theories that aim to
  characterize the same thing, we can think of the
  items/structures they each propose as
  counterparts.
• So here were concerned to identify the
  counterparts of sensations as states of
  perceivers, in the scientific image.
• So where are these sensa? In the brain (i.e.
  where the relevant physical events in the brain
  are occurring). When I seem to see a red
  triangle, a red-triangular sensum is occurring
  in some (visual) region of my brain.
• They dont seem to be in the brain, sure but
  then, they dont seem at all we dont perceive
  brain events in propria persona, in fact, we
  dont really perceive them at all these things
  are awarenesses (or something like that) not
  items we are aware of.
• Are we stuck with the primitive predicates of the
  MI? Clearly not. Form must have content, but
  content comes in other varieties than the
  familiar ones.

42
More on the SI view of ourselves

• Theories dont get their meanings from the
  observation language they get meanings from
  applications use. So the bridge laws (see
  above) dont work as partial definitions, leaving
  the rest of the theorys content up for grabs.
  They merely coordinate different conceptual
  frameworks, and allow us to use our grasp of one
  to learn to apply/use the other.
• The qualities of sensa may be found among the
  contents of a scientific description of states
  that occur in the brains of organisms like us.
• The scientific image converts us into complexes,
  multiplicities, even though we think of ourselves
  as individuals. All the descriptive facts about
  us must be re-framed to fit with this logical
  shift the normative view of things is something
  different, though.