CSCL 2009 Paper

Case Study #3 of Group Cognition Candy in the last year Chil no no no Candy er the year before next Collaborative organization of cognition Talk & gesture distribute across people & media Indexical network of group cognitive processes Cognition as a public social practice

Core Research Questions for CSCL CSCL is “ centrally concerned with meaning making ” (Koschmann, 2002) Research question #1 : how does meaning making unfold in CSCL small groups? Research question #2 : how do online groups achieve cognitive accomplishments? Research question #3 : how should CSCL environments & activities be designed to support group meaning making and group cognition?

Designing Support for Groups Single-user productivity tools: known technology & design methods (HCI) Social networking: new technology & haphazard design (Web 2.0 user-driven) Software of group learning: CSCL (computer-supported collaborative learning) Much more complex HCI & design & dissemination issues Much progress in virtual community social computing, but not in small-group collaboration in past 20 years

3 Levels of Cognitive Description The individual actor (person) is described by (various theories in) cognitive psychology The small-group cognition is what emerges in the interactions among the utterances of the individual participants The social / cultural / community of practice / linguistic community is the institutionalized, persistent, shared results of the above The study of these different levels requires different Units of Analysis

Science of Groups: Unit of Analysis Sciences of individual cognition Social psychology Educational psychology Cognitive psychology Organizational management Sciences of social & cultural cognition sociology, cultural anthropology, linguistics, etc. But no sciences of small-group cognition!

Science of Groups & Design HCI was based on theories of cognitive psychology of the individual CSCW was based on social theories of communities and large organizations Now we need a science of the small group Especially the computer-mediated, online, virtual group To guide design of instructional technology for collaborative learning, we need to understand how online small groups build knowledge, solve problems, pursue inquiry, explore, conceive, design

Sciences of Groups Of course, there has been research and theories about small groups What are the advantages and dangers? Do students learn more in groups? What about group size, heterogeneity, gender, socio-economics, …? But not about how groups achieve cognitive goals

Sciences of Groups Lots of talk of “meaning making” and “sense making”, but no analysis of how this takes place My claim: we can observe the interactional co-construction of group meaning in small groups (Meaning making and learning cannot be observed in individual or community cognition – making these contested and mysterious)

“Cognition” Psychology: group cognition is distortions of individual cognition (“group think”, “mass”) AI: cognition is computation (by any substrate) Distributed cognition: individual extended by artifacts and external memories Post-cognitive theories: cognition does not just take place in individual human heads

Cognition in CSCL CSCL is centrally concerned with the cognition of small groups in online sites Cognition at the small-group unit of description Cognition mediated by designed artifacts Analysis at the small-group unit of analysis Post-cognitive theory

Post-Cognitive Theories Philosophic basis in Hegel, Marx, Heidegger, Wittgenstein, Merleau-Ponty Social science basis in Giddens, Bourdieu, Geertz Analytic basis in Ethnomethodology and Conversation Analysis: Garfinkel, Sachs, Schegloff, Livingston

Post-Cognitive Theories Winograd & Flores, Dreyfus, Schön, Polanyi: critique of cognitivism Suchman: situated action Lave & Wenger: situated learning Engeström: activity theory Hutchins: distributed cognition Latour: actor-network theory Scardamalia & Bereiter: knowledge building

Post-Cognitive Theories “ Cognitive processes may be distributed across the members of a social group .” Hutchins (2000, p. 176) “ The cognitive properties of groups are produced by interaction between structures internal to individuals and structures external to individuals .” Hutchins (1995, p. 262) “ The group performing the cognitive task may have cognitive properties that differ from the cognitive properties of any individual .” Hutchins (1995, p. 176) But he analyzes socio-technical systems and the cognitive role of highly developed artifacts (airplane cockpits, ship navigation tools) These artifacts have encapsulated past cultural knowledge (community cognition) But not the cognitive meaning making of the group itself

Other Theories of Groups Provide some nice studies of the pivotal role of small groups, but do not account for this theoretically They are based on either a psychological view of individuals or a sociological view of rules, etc. at the community level None of them have a foundational conception of small groups as a distinct level They confuse talk of group level and social level They lack an account of the relationships between individual, group and community

Preliminary Explorations of Groups Social psychology (but reductionist) Organizational management (ditto) Teasley & Roschelle (1993) , Schwartz (1995), Cohen (2002), Barron (2003), Nosek (2004) Koschmann on practices of understanding (medical PBL) Suthers on intersubjective up-take (asynch.) Stahl (2006) Group Cognition: Computer Support for Building Collaborative Knowledge , MIT Press Stahl (2009) Studying Virtual Math Teams , Springer

The Concept of Group Cognition Group cognition (my working hypothesis): cognitive processes can arise through the interactions within a small group of participants — not just externalization of individual mental representations, but emergent result of situated interaction Group cognition is not a matter of a physical group with a brain or persistent presence It is a matter of meaning making through the interaction of semantic artifacts (words, drawings, symbols, documents) situated in a structured network of other meaningful artifacts (situation, world, group context, indexical field)

The Concept of Group Cognition E.g., we can observe group cognition in a years old chat log — in the meaning making of the chat postings in the physical absence of any group participants The postings are read as meaningfully designed by humans to interact with other human postings But the cognitive accomplishments (e.g., problem solving) are in the interactional discourse among the textual postings, not the physical interaction of the people.

The Concept of Group Cognition The group as actor and group cognition are not physical objects or mental objects, but theoretical constructs resulting from analysis at the group level of description (like cultural norms and social rules at the social level) E.g., interpersonal trains of thought, shared understandings of diagrams, joint problem conceptualizations, common references, coordination of problem-solving efforts

The Concept of Group Cognition The group and its cognitive accomplishments are enacted in situated interaction The cognitive accomplishments emerge from the network of meaningful references built up by the individual textual postings in chat E.g., planning, deducing, designing, describing, problem solving, explaining, defining, generalizing, representing, remembering and reflecting as a group

Mediation by Small Groups The small-group level of description has an epistemological, hermeneutic, pragmatic, semantic and scientific priority “ Small groups are the engines of knowledge building. The knowing that groups build up in manifold forms is what becomes internalized by their members as individual learning and externalized in their communities as certifiable knowledge” [ Group Cognition , p. 16].

Toward a Science of Virtual Groups Summary of argument : When small groups engage in cooperative problem solving or collaborative knowledge building, there are distinctive processes of interest at the individual , small-group and community levels of description, which interact strongly with each other. The small-group level has no corresponding science A post-cognitive science of virtual groups is particularly needed and possible

How to Build a Science: 5 Steps Define the domain of the science Explore the domain Capture a data corpus Select, adapt, refine and master methods for analyzing the data Organize analytic findings in a framework of theoretical conceptualizations

VMT as a Model of a Science Design-based research: iterative cycles Spring Fest 2005, 2006, 2007 and others by collaborators, in my courses, misc trials Over 2,000 student-hours of data (576 sessions) Almost 200 academic research publications Preliminary explorations: Group Cognition; Early Studies: Studying Virtual Math Teams

1. Create the Domain Design-based research evolves the technology with the pedagogy, methods of analysis, usage feedback from data analysis and theory From off-the-shelf AOL Instant Messenger to VMT From Math Forum “problem-of-the-week” to four-hour open-ended math mini-world From one-shot chats to Spring Fest sessions to mini-curricula

3. Establish a Data Corpus 2,000 student-hours of naturalistic usage Variety of scenarios: math problems, ages, group sizes, lengths of sessions, technologies Detailed logs Replayer to view and study interactions No data on individual factors or cultural Capture everything that entered into the interaction and was shared in the group interaction — available in detail in data

4. Analyze the Data Inspired by conversation analysis and ethnomethodology Use the replayer Select excerpts of interest Threading analysis Identify methods of group interaction Identify group cognitive achievements

5. Some Initial Findings A “simplest systematics” of conversation in VMT (Zemel, SVMT , 2009) – how text chat differs from speech Reading’s work in VMT (Zemel, Cakir & Stahl, CSCL, 2009) Recreate adjacency-pair structure as threading in chat (Zemel, Xhafa, Cakir, SVMT , 2009) Participant design across media (Cakir, SVMT , 2009)

5. Some Initial Findings Coordination of text & graphics in dual-interaction space (Cakir, Zemel & Stahl, CSCL, 2009) Grounding thru interactional organization establishing an “indexical ground of reference” (Hanks, 1992) as a network of situated references ( ijCSCL 4 ( 2 ))

5. Some Initial Findings The scientific study of intersubjectivity, common ground, shared understanding, sequentiality, indexicality can be done using ethnomethodologically inspired conversation analysis (Garfinkel, Sachs, Schegloff) “ Everything that can be counted does not necessarily count; Everything that counts cannot necessarily be counted ” – Albert Einstein

5. Some Initial Findings Problem solving discourse is driven by proposal/response interactions (Stahl, RPTEL , 2006) Groups construct a joint problem space through interactions that involve temporality, positioning and knowledge artifacts (Sarmiento, ICLS, 2008) VMT participants intricately coordinate visual, narrative & symbolic reasoning/inscriptions (Cakir, SVMT , 2009) Information questioning proceeds through interaction to elaborate what is sought (Zhou, SVMT , 2009) Groups construct an indexical field that lends contextual meaning to elliptical utterances (Stahl, Koschmann, Zemel, ICCE, 2009)

5. Some More Initial Findings (Medina, Suthers, Vatrapu, SVMT , 2009) uptake analysis of methods in VMT (Koschmann, Stahl & Zemel, CSCL, 2009) recipient design of proof (Wee & Looi, SVMT , 2009) model of threading (Powell, et al., SVMT , 2009) co-construction of math reasoning (Fuks, et al., SVMT , 2009) avoiding chat confusion (Trausan, et al., SVMT , 2009) polyphony of discourse (Rose, et al., SVMT , 2009) agent support for VMT (Shumar & Charles, SVMT , 2009) group agency

Some Initial Theory Important cognitive processes occur distinctively on individual, group and community levels of description They are appropriately studied at the corresponding unit of analysis The levels influence each other, but are not reducible to each other Often, group cognition can be best observed, because it takes place publically and explicitly and has not yet been reified or institutionalized

Scientific Issues A rigorous science can take many forms — e.g., predictive mathematical physics vs. case-based descriptive history — but it is generally concerned with issues of: Objectivity Reliability Generalizability

a) Objectivity The data is automatically logged No selective perspective (camera angles, lighting, choice of heard or remembered) No interpretive transcription Logs include all relevant details of interaction Replayer allows extremely detailed analysis: like a digital video with timecode, slow-mo, pause, next/previous action

Objectivity The VMT Replayer displays everything that was shared by the participants – exactly the group unit of analysis Everything that was shared went thru server Log includes every action with exact time Replayer is created exactly like original displays were created Analyst accesses exactly what would have been available to a “fly on the wall” – but with ability to proceed as needed and repeat

b) Reliability Data sessions with multiple analysts Using logs, replayer Discuss individual chat postings & moves Stricter standard than inter-rater reliability

c) Generalizability Analyst group has experience with many chat and classroom math interactions Conversation analysis argues that there are necessarily general methods people use Members of a linguistic community share a necessarily limited set of recognizable and identifiable methods for accomplishing everyday interaction tasks Ethnomethodology argues that utterances and other interactive moves are “accountable” The way they are organized displays to others the general methods, to recognize them as what they are

The Power of this Science Analysis of group interaction (group cognition) Can show HOW (not just quantitatively THAT) The group can achieve cognitive accomplishments that none of the individual members of the group can (the group accomplishment may not even be completely understood by individuals)

Vygotsky’s Zone of Development Analysis of small-group interaction can EXPLAIN the ZpD by showing how a group-level “train of thought” is built by contributions of many individuals And by latent relationships implicit in the semantic relationships — never present in any individual mind. The meanings are human in origin, design and nature, but not necessarily configured by any individual the way they are by the group — although there is subsequently the possibility for individuals to take them up as their own resources for individual cognition

Summary There is a scientific lacuna between sciences of the individual and sciences of communities There are important cognitive achievements at the small-group level of description These should be studied by a science of groups Online small groups are becoming increasingly possible and important in the global networked world A cognitive science of virtual groups could help the design of collaborative software for working and learning

Core Research Questions for CSCL Post-cognitive design-based research such as the VMT project analyzing interaction can build a science addressing: Research question #1 : how does meaning making unfold in CSCL groups? Research question #2 : how do online groups achieve cognitive accomplishments? Research question #3 : how should CSCL environments & activities be designed to support group meaning making and group cognition?

For Further Information Slides: http://GerryStahl.net/pub/cscl2009ppt.pdf Website: http://GerryStahl.net Email: [email_address] Group Cognition (2006, MIT Press) Studying Virtual Math Teams (2009, Springer)

CSCL 2009 Paper

More Related Content

What's hot (20)

Viewers also liked (6)

Similar to CSCL 2009 Paper (20)

Recently uploaded (20)

CSCL 2009 Paper