Title: Confabulation Theory: The Mechanism of Thought
1Confabulation TheoryThe Mechanism of Thought
2Mental World
3What is Cognition?
- Cognition
- Understanding and trying to make sense of the
world - Information processing
- Development of concepts
- The mental functions, mental processes and states
of intelligent entities with focus on - comprehension,
- inference,
- decision-making,
- planning and
- learning etc.
4Cognition
- Cerebral Cortex
- Seat for cognition
- How is Cerebral cortex able to do what it is able
to do? - Fuzzy explanations
- No clear cut perspective in exact terms
- Introducing Confabulation Theory
- A discrete insight into possible mechanism of
thought
5Confabulation Theory
- Proposed by Robert Hecht-Nielson
- All human cognition and behavior based on one
simple, non-algorithmic procedure that has been
named confabulation - All aspects of cognition carried out using
- a single type of knowledge and
- a single information processing operation called
(confabulation)
6Relevance
- Radical novel approach
- First-of-its-kind concrete model
- Issuing deeper insights into process of cognition
- Augmenting present
- Artificial intelligence
- knowledge discovery
- knowledge management
7Roadmap
- Cognitive World Object Representation
- Knowledge Links
- Confabulation
- The Mathematics of Cogent Confabulation
- The Origin of Behaviour
- Conclusion
8Cognitive World Object Representation
- Cerebral Cortex
- 4000 discrete, localized, disjoint patches
- Thalamocortical Module
- cortical patch and first-order thalamic zone
uniquely paired by reciprocal axonal
interconnection - Module
- one attribute of an object of mental world
9Cognitive World Object Representation
10Cognitive World Object Representation
- How are these modules used?
- Make groups of 60 symbols
- Module may consist of thousands of neurons
- Symbol one possible descriptor of an attribute
- One neuron may belong to more than one symbols
- Symbols have to be permanent
11Cognitive World Object Representation
12What is Knowledge?
- Knowledge accumulates in discrete units
- Knowledge link or knowledge unit is an axonal
linkage between source symbol and target symbol - Source and target symbols belong to different
modules
13Knowledge Links
14Formation of knowledge links
- Two symbols become co-active to send out signals
- Synapses are strengthened in the process of
learning - Permanent strengthening during sleep
- Billions of knowledge links
- Humans and animals are 'smart
15Confabulation
- One and only one information processing operation
- Localized
- Winners-take-all
- Symbol with highest total knowledge link input is
conclusion of confabulation
16Implications of Knowledge Link
- Source symbol neurons send signals to millions of
transponder neurons - Only few thousand transponder neurons become
highly excited - 10 of the target neurons receive signals from
multiple transponder neurons
17Implications of Knowledge Link
18Confabulation Neuron Level
19Cogency
20Cogency
- Aristotelian model An appealing model for
cognition p(eaß?d) - Wrong model
- Alternate model Cogency p(aß?de)
- a,ß,?,d Assumed facts
- e Conclusion
21Cogency Theorems
- Thm 1 If aß?d gt e exclusively, then
maximization of cogency produces one and only one
answer e - Aristotelian logic information environment
maximizing cogency gives logical answers
22Cogency Theorems
- Cogency calculation only in trivial situations
- Confabulation Product p(ae).p(ße).p(?e).p(de
) - Thm 2 p(aß?de)4
- p(aß?de)/p(ae).p(aß?de)/p(ße).
p(aß?de)/p(?e).p(aß?de)/p(de).
p(ae).p(ße).p(?e).p(de) - In non-exceptional cases, p(aß?de)4Cp(ae).p
(ße).p(?e).p(de)
23Confabulation Examples
- Some examples from test
- She could determine (whether, exactly, if, why)
- If it was not (immediately, clear, enough, true)
- Earthquake activity was centered
- For lack of a (unified, blockbuster,
comprehensive, definitive) - A lack of (urgency, oxygen, understanding)
- Regardless of expected outcome, length
- Automatic emergence of semantics and grammar
24Why Cogency and not Bayes Law?
- p(?)0.01 p(e)0.0001 p(aß?d?)0.01
p(aß?de)0.2 - p(aß?de) 20 p(aß?d?)
- p(? aß?d) 5 p(eaß?d)
- Bayes Law gt ?
- Cogency gt e
25Quiz! (For those who are sleeping)
- Quickly select a next word for each of the
following - Company rules forbid taking
- Mickey and Minnie were
- Capitol hill observers are
- Paper is made from
- Riding the carousel was
26Why Cogency and not Bayes Law?
- Typical answers
- Naps
- Happy
- Wondering
- Wood
- Fun
- the also viable Bayes law
27Conclusion-Action Principle Origin of Behavior
28Conclusion-Action Principle Origin of Behavior
- Every time a confabulation operation on a module
reaches a conclusion, an associated set of action
commands are launched - Winner of a confabulation competition employs
skill knowledge to launch an action - Skill knowledge, and skill learning are not parts
of cognitive thinking
29Hindering Blocks
- Conceiving and then precisely defining a
confabulation architecture - Conceiving and then precisely defining the
thought processes - Conceiving and executing an appropriately staged
sequence of learning opportunities
30Conclusion
- A new dimension to the mechanism of thought
- Based on Cogency (refutes Bayesian model)
- If proved correct
- Better insight into human cognition
- Will redefine the outlook towards various AI
problems - Radical improvement in present ML techniques
31References
- Robert Hecht-Nielson, Cogent Confabulation,
Neural Networks Letter, 2004 - Robert Hecht-Nielson, Confabulation Theory A
Synopsis, Institute for Neural Computation, 2005 - Robert Hecht-Nielson, The Mechanism of Thought,
International Joint Conference on Neural
Networks, 2006 - scholarpedia.org/confabulation
32ThanQ
- Animal cognition maximizes cogency, and in a
non-logic environment, cogency maximization
implements what I call the duck test." - Robert
Hecht-Nielson - There must be some event that triggers every
behavioral event, and it had to be the same in
every instance, whether we're thinking, moving,
or whatever." - Robert Hecht-Nielson