psych 421
deefer
Chapter8.4_Hybridarchitectures-revisedfor3rdedition.pptChapter 8.4:
Hybrid architectures
*
Cognitive Science José Luis Bermúdez / Cambridge University Press 2020
Two approaches to mental architectures
Physical symbol systems
Artificial neural networks
- These need not be mutually exclusive
- One prominent attempt at integrating both approaches is John Anderson’s ACT-R/PM cognitive architecture
Cognitive Science José Luis Bermúdez / Cambridge University Press 2020
Cognitive Science José Luis Bermúdez / Cambridge University Press 2020
ACT-R/PM (Adaptive Control of Thought—Rational/Perceptual Motor)
- ACT-R is modular
- Cognitive modules can only access sensory information through buffers
Cognitive Science José Luis Bermúdez / Cambridge University Press 2020
Cognitive Science José Luis Bermúdez / Cambridge University Press 2020
ACT-R architecture
Cognitive Science José Luis Bermúdez / Cambridge University Press 2020
Cognitive Science José Luis Bermúdez / Cambridge University Press 2020
ACT-R/PM (Adaptive Control of Thought—Rational/Perceptual Motor)
- Two types of cognitive modules: declarative and procedural
Declarative (knowledge that): e.g., knowing that Paris is the capital of France
Procedural (knowledge how): e.g., knowing how to speak French
- Declarative knowledge is achieved though chunking
- Procedural knowledge is achieved through production rules
Cognitive Science José Luis Bermúdez / Cambridge University Press 2020
Cognitive Science José Luis Bermúdez / Cambridge University Press 2020
What makes it a hybrid architecture?
There is no central processor
Decisions are made subsymbolically, prior to cognitive modular processing
Serial processing: only one production rule can be active at a time
Pattern-matching module chooses which production rule to be active
Pattern-matching module chooses according to utility—which production rule will achieve the system’s goals most efficiently
Cognitive Science José Luis Bermúdez / Cambridge University Press 2020
Cognitive Science José Luis Bermúdez / Cambridge University Press 2020
Subsymbolic processing
Each declarative chunk and each production rule is symbolic, but the information that determines their utility is subsymbolic.
Subsymbolic processes determine buffer placement and activation level, which determine whether a chunk or rule is used.
Cognitive Science José Luis Bermúdez / Cambridge University Press 2020
Cognitive Science José Luis Bermúdez / Cambridge University Press 2020
Subsymbolic processing
Cognitive Science José Luis Bermúdez / Cambridge University Press 2020
|
Table 10.2 Comparing the symbolic and subsymbolic dimensions of knowledge representation in the hybrid ACT-R/PM architecture |
||||
|
|
Performance Mechanisms |
Learning Mechanisms |
||
|
|
Symbolic |
Subsymbolic |
Symbolic |
Subsymbolic |
|
Declarative chunks |
Knowledge usually facts) that can be directly verbalized |
Relative activation of declarative chunks affects retrieval |
Adding new declarative chunks to the set |
Changing activation of declarative chunks and changing strength of links between chunks |
|
Production rules |
Knowledge for taking particular actions in particular situations |
Relative utility of production rules affects choice |
Adding new production rules to the set |
Changing utility of production rules |
Cognitive Science José Luis Bermúdez / Cambridge University Press 2020
Take home message
- The symbolic, physical symbol system approach can be integrated with the subsymbolic, artificial neural network approach
- Some tasks might require one architecture and some might require the other
Cognitive Science José Luis Bermúdez / Cambridge University Press 2020
