ANTH245 2007-10-01

← Go to Main Page

Business

• The Midterm is available on Blackboard
• Please send in your responses on time

Recap

• What is Coding?
• The process of converting texts into codes
  • A kind of transduction
  • A kind of reductionism
  • informal → formal
• A : B :: Text : Code :: Discourse : Ontology
• The reduction of A to B is structuralism

Segue

• Structuralism = Cognitive Science
• Just as information culture represents text as code, CogSci represents activity as plan
• Activity = ritual, work, practice = parole, event

Intro

• We are now officially turning the tables -- from "culture as information" to "information as culture"
• We will explore continuities and discontinuities between the two approaches
• Path:
  1. Computer as metaphor
  2. Computer as tool
  3. Computer as artifact
  4. Next week: Expanding the scope

Suchman 1983

Background

• Description of Lucy Suchman
• One of the pioneers of "corporate anthropology," but from a critical perspective
• Sets the method:
  1. Office as ethnographic site -- "village"
  2. Computer as core artifact -- the "churinga"
  3. Work as ritual
  4. What are the symbols and myths?
     • discourse, planning tools, documents
     • Management and Labor

Procedures

• What is the problem? The status of formal procedures ...
  • The "natives" view procedures as a description of work
  • But the anthropologist sees things otherwise
  • Note that here the denial of the native point of view does not strike as ethically problematic (as it did with Rappaport)
• What are the "real" procedures?
  • The description derived from looking at the transcripts and other empirical data
• So ... two kinds of description
  • A: Formal -- Standard Procedures; flowchart
  • B: Informal -- Ethnographic (thick) description; narrative
• Comparisons:

Author	A (Formal)	B (Informal)
Bateson	Purposive Mind	Greater Mind
Rappaport	Cognitive Models	Environment → Operational Models
Levi-Strauss	Myths → Structures	Event
Colby	Folktales → Templates	Experience
Suchman	Plans and Procedures	Practical action

Suchman 1988

• Ethnomethodology
• Method: to study representations of work and the work of representations

AI

• Compare to behaviorism
• Considers mentalist concepts -- memory, thinking, etc. -- without introspection
  • Instead, the computer serves as an objective model for thinking
  • Brain = Computer, Mind = disembodied Pattern
• Difference between organic and silicon substrates not relevant

The Turing Test and ELIZA

• Describe the test
• Describe ELIZA
• ELIZA does not really pass the test, but it illustrates that communication depends upon the listener imputes intentions to the speaker, even when there are none
  • Communication depends upon the "documentary method," where utterances are seen as indices, or documents, of inner mental states.
  • For AI these inner states are plans

Plans and Actions

• For AI, plans adequately describe actions
• But plans can't really do that
  • Actions are more complicated than plans
  • Plans orient people before they take action (e.g. canoing the rapids)
  • Plans are convenient to describe the results of actions after the fact
  • (The AI concept of plan elides the differences between conscious and unconscious thought as described by Bateson)

Human-Machine interaction

• H-M interaction is asymmetric: humans bringer greater "conversational resources" to the table
  • Humans read context
  • Inference of intent in conversation seem context dependent; more than matching plans with situations

Whiteboards and the "work of representation"

• ("work of representation" not Suchman's phrase)
• Whiteboard as "cultural model" (Colby)
• Artifacts like the whiteboard help construct conversations; they do not simply capture the information content of a meeting (for example)

Forsythe 1993

Background

• Influenced by Suchman
• Sadly died in 1997 at 50

Technology as a Cultural System

• Traditional cultural anthro; Clifford Geertz
• Not a practicioner of ethnomethodology
• Culture as "tacit knowledge" and that which is taken for granted

Subject: Knowledge Engineers

• Cognitive Science → AI → Expert Systems (ESs)
• Content of ESs acquired by "knowledge engineers" (KEs)
• KEs convert human knowledge to machine "knowledge"
  • Compare to the "coding" of Colby, HRAF

Falling off the knowledge cliff

• ESs have precipitous shortcomings, e.g. lack of common sense

What counts as "work"

• Forsythe found that KEs only regard actual programming as "real"
• Everything else -- from talking about epistemology to all social interaction -- is viewed as at best secondary and at worst irrelevant to the work of building ESs
• She calls this "deleting the social"

Effect of "deleting the social" on design

• Forsythe found that KEs project this narrow understanding of work onto the domains (e.g. medicine) for which they build expert systems
• They select one expert and interview this person in isolation from actual practice
• Thus, no reference to socially distributed knowledge or practical action
• She argues that the absence of this kind of knowledge probably accounts significantly for ESs falling off the knowledge cliff