Development is a layering process. New behaviors are layered on top of old. The
progression is linear and predictable at the beginning (Piaget).
Early learning goes through a progression of sensori-motor to concrete to (sometimes)
abstract.
Its therefore difficult to learn something new until the needed foundation skills are
available. Hence there is a zone of proximal development (Vygotsky) in which children
learn most effectively.
Learning becomes less linear (branched) as the child grows older. Knowledge acquisition
is less predictable, both in time and order.
Skill acquisition goes from conscious to unconscious (Piaget) or in activity theory
language, from actions to operations.
Language plays a central role in learning (Vygotsky), because learning is a social
activity. Understanding text can be viewed as reconciliation of writers,
readers and other voices (Bakhtin).
Learning proceeds at societal scale (socio-cultural learning) as well as at the
individual or organizational level.
Studying development is important for understanding behavior:
Much knowledge work is about creation and consolidation of knowledge (a kind of
learning) and is sometimes called "organizational learning".
Psychological theories of development show us that to understand behavior, you need
to understand the evolution of a behavior on top of others. You cannot understand it by
obervation or rational agent theories, which ignore history.
Cognitive Science
Cognitive science is a collection of perspectives on human thought, rather than a single
theory.
There is structure in dialogue at the level of users intent (speech act theory).
Cognitive models of sensori-motor behavior work very well for predicting behavior on
low-level tasks (GOMS and EPIC).
Choice of representation can make or break problem-solving. Representations dont
need to be "in the head". People often use things in the world as part of their
cognitive processes.
Metaphors provide a wide-spread shared basis for language and thought. They allow the
building of new concepts on old ones.
Simple pattern-recognition schemes (LSA) can perform surprisingly well at
natural-language tasks. Meaning in these systems depends on context and other texts, which
many linguists have argued is true of human language.
Personality and Emotion
Personality and emotion are amenable to quantitative analysis. The data can come from
other peoples assessment, from self-assessment, or from secondary cues (use of
language).
Personality and emotion can be modeled with low-dimensional linear models. Compared to
high-level cognition, they are simple processes. However, their coupling with high-level
processes (Ekman and Friesan) means that both felt emotion and emotional displays are
affected by the individuals thought.
Much emotion and personality information is conveyed through non-verbal cues.
Groups
Group processes are complex. Communication is often a problem, so groups are rarely the
sum of their parts. Choice of media is important to support group tasks. Rich media for
creative and open-ended tasks, and low-richness media for specific or routine tasks.
Face-to-face groups tend to be centralized and are prone to group-think. Email groups
are more egalitarian and have more divergent thought. But F2F is important for group
consensus and coherent action.
Activity theory provides a simple but universal structure for human action. It has a
hierarchy of behavior, acknowledges development as part of activity, and is applicable to
individuals or groups.
Disciplines have different working vocabularies, which make interdisciplinary work hard.
Mobile representations are one solution to communication across time, space and across
discipline.
Social Networks
Interpersonal relationships are amenable to quantitative (graph) analysis.
There are universal structures in social networks such as centrality (which reifies the
informal notion used in the group papers above) and role and position. They are not
properties of an individual, but of the individuals position in the network.
Networks mediate dynamic behaviors such as diffusion of innovations. Diffusion happens
as a bell-curve histogram of adoption time. Heterophily is important for innovations to
move across demographic boundaries.
Design and Knowledge Creation
Ethnography is one approach to understanding context, based on observation only.
Ethnomethodology is an activist approach, which tries to extract user models of the task.
Contextual design involves careful contextual inquiry, and then design at a
"suitable distance" to allow creativity maintaining user involvement.
Applications are often used in "toolbelts" that evolve along with practice.
Fluidity from one application to another, and the ability of an application to evolve, are
very important.
Knowledge creation can be viewed as a cycle of explicit-tacit knowledge conversion. This
cycle can be reconciled with activity theory.