Connective Knowledge

The Recognition Factor (connectivism conference abstract)

By Stephen Downes
Connective knowledge is based on pattern recognition of emergent phenomena in networks. In order for a pattern to have any meaning, therefore, it must be recognized. This means that knowledge formation in a connective environment is a combination of two elements: the perception, which is the pattern to be recognized, and the perceiver, who does the recognizing. Knowledge, therefore, is not uniquely inherent in a network, but exists only insofar as it is recognized to exist. This talk will explore this argument and its implications on a theory of connective knowledge.

Stephen Downes papers:

Learning networks and connective knowledge (October 2006)

The Buntine Oration: Learning Networks (October 2004)

An Introduction to Connective Knowledge (December 2005)

a. Types of Knowledge
three types - qualitative (Greeks), quantitative (Arabic Renaissance), distributed (connective, interactive)
Probabilist knowledge is quantitative, not distributed
Connective knowledge requires interaction, is knowledge of the connection

b. Interpretation
We do not obtain direct knowledge about the world but rather interpret what we receive through our senses
"the apple is red"
- depends on the light
- depends on our organisation of the spectrum
- depends on our prior knowledge and language

c. Emergence
is fundamentally the result of interpretation applied to connective knowledge
- a wave effect emerges from falling dominoes
- a person emerges from pixels on a TV screen

d. Physicality

physicality is useful for sorting b/w "correct" (reality) and "misperceptions" (mirage)
there is nothing in our interpretation that is inherently based on physical reality
non physical entities have properties that are useful, eg. "purple prose"
Our perceptions are real but properties like "red" and numbers like four are not real
Our interpretation of connections is distinct from the actual set of interactions in the world
eg. conspiracy theories may have no real world basis but be real to particular people

comment: isn't the theory practice spiral useful?

e. Salience and inference

f. Associationism
Hume and Mill
associationism - two things that are similar become associated in the mind
(if a tiger tries to eat us then we will think that another tiger will try too)

Hebbian associationism - if two neurons fire at the same time a connection will tend to form b/w them
Boltzmann associationism - ?

g. Distribution
A political party is a distributed entity

h. Meaning
properties (eg. redness) can exist as abstractions from physical entities
"redness" is an example of distributed meaning - we can't point to a "redness" concept location in our mind
"redness" arises from a society of neurons

"paris" has different meanings
to understand "paris" requires understanding many things including the concept of naming things

i. Shared Meaning
Wittgenstein: meanings only exist when shared by a community of speakers

cf empirical truth
"snow is white if and only if snow is white"
ie. an independent existing physical world

different languages interpret the world differently (not withstanding Chomsky - mentalese)
Wittgenstein: "meaning is use" - shared meaning of "Paris" is emergent property of interactions between people using the word
it's all interpretation!
knowing isn't possessing instances / facts but is having a pattern of neural activity, a habit of mind (Hume)
young children learn a language without knowing it in a cognitive sense - not innate (Chomsky) but a learnt organisation of neural connections

language is both personal and social
language influences speaker and speaker influences language

j. Organisation
language is both social and mental
social: meaning arises through shared language
mental: meaning arises through neuron structures
network study (Minsky, Papert mentioned ... )
any network can come to know just like a human mind can come to know

k. Social Knowledge
connected knowledge b/w different members of society
eg. how to fly a person from England to Canada in a 747
eg, the market value of wheat

social knowledge
Surowiecki: The wisdom of crowds

l. Power laws / inequalities
o. Knowing
networked knowledge may be 'good' or 'bad'
cascade phenomenon:


t.Knowing networks

These are the criteria for a knowing network

u. Remnants
networked knowledge is liable to error because it is dependent on interpretation (partial patterns)
historical problem of one man or small group distorting media for their own ends