CARDIFF UNIVERSITY UK Cardiff School of Social Sciences

1
CARDIFF UNIVERSITY UKCardiff School of Social
Sciences
www.cf.ac.uk/socsi/expertise
2
Phases of work

• SSK informed by methodological relativism

3
University of Chicago Press1985/1992
4
Golems
5
Phases of work 2

• Critique of AI based on ideas of tacit
  knowledge/the social (Collins Kusch)
• Work on expertise motivated by The South African
  Question
• THE PERIODIC TABLE OF EXPERTISES
• (based on tacit knowledge)

6
Work on Expertise by Harry Collins and Robert
Evans

• The Third Wave of Science Studies Studies of
  Expertise and Experience'
• Social Studies of Science, 32, 2, (2002) 235-296

Rethinking Expertise University of Chicago
Press August 2007
7
TACIT KNOWLEDGEContinue the series

• 2, 4, 6, 8,

8
Possible continuations

• 10 (2,4,6,8,10,)
• 2 (2,4,6,8,2,4,6,8,)
• 8 (2,4,6,8,8,4,6,2,2,4,6,8,)
• 4 (2,4,6,8,4,6,8,10,6,8,10,12,)
• 6 (2,4,6,8,6,8,10,12,10,12,14,18,)
• 1 (2,4,6,8,1,3,5,7,-1,1,3,5,)
• 3 (2,4,6,8,3,5,7,9,4,6,8,10,)
• 5 (2,4,6,8,5,7,9,11,...)

9
Another continuation

• Who (2,4,6,8,who do we appreciate?)

10
Ubiquitous and Specialist Tacit Knowledge

• Natural language speaking, keeping your distance
  on the pavement, everything you need to live in
  society
• The things that people tried to put into
  computerised expert systems. (In my case the
  specialist tacit knowledge of gravitational wave
  physics) www.cf.ac.uk/socsi/gravwave

11
Periodic Table of Expertises
12
Interactional Expertise
13
Interactional Expertise

• Can you learn a domain language without the
  domain abilities?

14
The Strong Interactional Hypothesis

• A PERSON WITH MAXIMAL INTERACTIONAL EXPERTISE AND
  NO CONTRIBUTORY EXPERTISE WILL BE
  INDISTINGUISHABLE FROM A PERSON WITH BOTH IN ANY
  TEST BASED ON VERBAL INTERCHANGE ALONE

15
The Imitation Game
Person with interactional expertise pretending to
have contributory expertise
Person with interactional and contributory
expertise
Judge (with contributory expertise)
16
Experimental configurations
17
Hypothesised Outcomes

• Pretender is Target Expected
• Expertise Outcome
• AColor-blind Color-perceiving Chance
• BColor-perceiving Color-blind Identify
• CPitch-perceiving Pitch-blind Chance
• DPitch-blind Pitch-perceiving Identify

18
Confidence levels

• 1 I have little or no idea who is who
• 2 I have some idea who is who
• but I am more unsure than sure
• --------------------------------------------------
  ----------------------------
• 3 I have a good idea who is who
• and I am more sure than unsure
• 4 I am pretty sure I know who is who

19
A colour blind judge guesses correctly at
confidence level 4
Judge's comment ... B's answer to the main
problems one faces is almost exactly what I would
say. ... The real clincher for me may strike a
colour normal person as odd the statement by A
that "I might not be able to enjoy things like
films". This seems a very strange idea, as I have
never seen colours normally, so can't see how
being colour-blind would affect my enjoyment of
them although perhaps this reflects each
person's personal outlook.
20
The Imitation Game Phase 2
?
?
21
Hypothesis proved!
22
Chance condition
23
Binary inspiral
24
An interferometer affected by a quadrupole
gravitational wave
25
Hanford and Livingston
26
University of Chicago Press 2004870 pages
27
Imitating a GW physicist
gwB
gwP
CHANCE ?
28
(No Transcript)
29

• NATURE
• 6 July 2006

30
GW scientists who preferred Collins

• ... I find that I lean to W. But Z is
  pretty darn good _ I'd be entirely unsurprised
  if you told me this was a control run and that
  you'd used responses from two experts.
• Set P looked more like they had been
  answered by looking up a book. Set Q looked
  as if they came rapidly out of the mind.

31
Different arrangements of GW imitation game

(GW) scientist (C)Ollins (L)ay person Non-GW
(S)cientist Astrophysicists and
Astronomers (E)vans
Chance Identify
I
I
C
I
C
I
32
Lay persons who preferred Collins

• I have no idea about the detail of any sets of
  answers, not knowing this field. I thought J
  was more persuasive as he/she seemed not to feel
  the need to elaborate on answers quite so much or
  set them in some wider didactic context. As such,
  J did not strike me as someone trying to
  persuade anyone else of their own credentials,
  presumably because they are not in question.
• My guess was based on accumulating evidence from
  the series of questions, rather than any
  particular one. It seemed to me that the
  responses of Q were going out of their way to
  appear knowledgeable and 'scientific/specialist'.
  I suspect that the specialists actually talk to
  one another in more natural terms as in J's
  answers, being able to assume shared background
  knowledge. I'm also aware, though, of how I'm
  interpreting responses to individual questions to
  fit in with my overall decision. As a possible
  get-out, of course individuals vary in manner -
  and a very senior scientist might give different
  kinds of answers to a junior one than to another
  senior colleague.

33
CONCLUSION

• YOU CAN LEARN A DOMAIN LANGUAGE WITHOUT THE
  DOMAIN ABILITIES !
• (interactional expertise without contributory
  expertise)

34
Other publications on expertise

• Collins, H. M. (2004) Interactional Expertise as
  a Third Kind of Knowledge' Phenomenology and the
  Cognitive Sciences, 3 (2) 125-143
• Collins, Harry, (Ed) (2008 forthcoming) Case
  Studies in Expertise and Experience special
  issue of Studies in History and Philosophy of
  Science, 39. 1 March

35
All will be found on the Website

• www.cardiff.ac.uk\socsi\expertise
• or Google
• Harry Collins Expertise

36
Philosophical Implications

• YOU CAN ACQUIRE A COMPLETE DOMAIN LANGUAGE, WITH
  ALL ITS TACIT COMPONENTS, WITHOUT ACQIRING THE
  EMBODIED TACIT KNOWLEDGE !

Many sociological and policy implications
What are the philosophical implications?
37
If a lion could talk
38
Social and minimal embodiment
Mom!
Grr!
39
Madeleine
she had never fed herself, used the toilet by
herself, or reached out to help herself, always
leaving it to others to help her' (p 58) spoke
freely indeed eloquently ... revealing herself to
be a high-spirited woman of exceptional
intelligence and literacy' (p56)
40
Interactional Expertise

• Minimal embodiment required to learn language
  (larynx, ears, brain)
• Minimal interaction with the physical world
• Cf the deaf, who have lots of interaction but
  have no ears, and struggle to learn the native
  language

41
It is social embedding that is crucial
WITTGENSTEINIAN AI
Autism Feral Children
Frontal Lobe Damage
42
A judge with perfect pitch guesses correctly at
confidence level 3
Here the judge said that hearing pitches in
peoples' voices was a valuable clue especially
since this was sometimes found to be annoying
whereas B didn't really seem to understand the
question
43
A judge with perfect pitch guesses correctly at
confidence level 4
Here the judge thought that only a person with
perfect pitch would use pitch to identify voices
and that this question alone definitely indicated
the person with perfect pitch. Participants
without perfect pitch would not be sure that
these things could be accomplished by a person
with perfect pitch and might think that replying
positively to the questions would be to fall into
a trap set by the judge.
44
A colour blind judge guesses correctly having
reached confidence level 3 at this point in the
questioning
Here the judge found the answers in the left
column referring to clothing implausible. He
found the embarrassment caused by drawing
something in the wrong colour evocative of his
own experience.
45
A colour blind judge guesses correctly at
confidence level 3
Judge's comment Participant A claims to have
trouble distinguishing "primary" colours, whereas
in my experience, it's the shades of colour that
present me with the most trouble also why red,
green and yellow?!
.
46
END