Web and Email Addresses

The IDA Model’s Cognitive
Cycle
Stan Franklin
Workshop on the Role of Consciousness in Memory
May 1,2004—FedEx Institute of Technology
The IDA Model
• Conceptual model of cognition
• Computational model plus
additional designed
but unimplemented features
• A uniform model of cognition
such as SOAR, Act-R, C&I
• Models a broad swath
of cognitive functions
IDA’s Cognitive Cycle
• Specifies the role of consciousness
in cognition
• Clarifies the relationship between
consciousness and the various memories
• Makes explicit the role of consciousness
in recruiting relevant resources
• Provides a tool for the fine-grained analysis
of various cognitive tasks
Perceiving
• Preconscious perception
• External or internal
• Processing of portions of the stimuli
• Construction of meaning
– Recognition
– Categorization
Store Percept in WM
• Percept stored in Working Memory’s
preconscious buffers
– Visuo-spatial sketchpad
– Phonological loop
• Buffers may contain earlier contents also
• Decay time measured in seconds
Local Associations
• Cued by the contents
of preconscious WM buffers
• Retrieves local associations from
– Transient episodic memory
– Declarative memory
• Contents of WM plus local associations occupy
– Baddeley’s episodic buffer
– Ericsson and Kintsch’s
Long-term Working Memory
Competition for
Consciousness
• Attention codelets view LTWM
• Form coalitions with
information codelets
• Vie to bring various portions
of contents to consciousness
• Coalition from a previous cycle
can win
• Factors include
– Relevancy
– Importance
– Urgency
– Insistence
– Recency
Conscious Broadcast
• Coalition with highest average activation is chosen
• Is said to be in the spotlight,
or to occupy the global workspace
• The information content of the coalition is
broadcast to all codelets
• GW theory postulates this broadcast
as the moment of phenomenal consciousness
Learning Occurs
• Memory updated using broadcast contents
– Perceptual memory updated
– Transient episodic memory updated
– Procedural memory updated
• Feelings and emotions modulate learning
• Declarative memories consolidated later off line
Setting Goal Context
Hierarchy
• Relevant behavior codelets
respond to broadcast
• Instantiate goal context hierarchy
(behavior stream) if needed
• Bind variables using information
from conscious broadcast
• Send environmental activation
to appropriate behaviors
Action Chosen
• Behaviors (goal contexts) get activation from
– Drives (feelings and emotions)
– Environment (external or internal)
– Other behaviors
• The single behavior is chosen that
– Is executable
– activation over threshold
– higher activation
than other such behaviors
Action Taken
• Chosen behavior binds variables
in its behavior codelets
• Then activates its behavior codelets
including at least one expectation codelet
• These behavior codelets
perform the task of the behavior
• This action may effect
the external or internal environment
Cognitive Cycle Processing
• Hypothesis — Like IDA’s, human cognitive processing is
via an iterating sequence of Cognitive Cycles
• Duration —
Each cognitive cycle takes roughly 200 ms
with steps 1 through 5 occupying about 80 ms
• Cascading — Several cycles may have parts
running simultaneously in parallel
• Seriality — Consciousness maintains serial order
and the illusion of continuity
• Start —
Cycle may start with action selection
instead of perception
Goals and Actions
• Conscious goal (intention) selection
– Consciously chosen over several cycles
• Consciously mediated actions
– Use the environment to inform the action
– Internal or external action
• Automatized actions
– Performed unconsciously
Example of Goal Selection
and Subsequent Actions
• Conscious goal selection
– Go to the fridge for orange juice
– Choice between go or wait, orange juice or coke
or water
• Consciously mediated action
– Find and grasp the handle
• Unconscious, automated actions
– Pull the refrigerator door open
Web and Email Addresses
• Stan Franklin
– [email protected]
– www.cs.memphis.edu/~franklin/
• ‘Conscious’ Software Research Group
– csrg.cs.memphis.edu/