Historical Introduction with a Focus on Parallel Distributed Processing Models Psychology 209 Stanford University Jan 7 2013 Early History of the Study of Human Mental Processes Introspectionism Wundt Titchener ID: 276412
Download Presentation The PPT/PDF document "Models of Cognitive Processes:" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Models of Cognitive Processes:Historical Introduction with a Focus on Parallel Distributed Processing Models
Psychology 209
Stanford University
Jan 7, 2013Slide2
Early History of the Study of Human Mental ProcessesIntrospectionism (Wundt, Titchener)
Thought as conscious content, but two problems:
Suggestibility
Gaps
Freud suggests that mental processes are not all conscious
Behaviorism (Watson, Skinner) eschews talk of mental processes altogetherSlide3
Early Computational Models of Human Cognition (1950-1980)
The computer contributes to the overthrow of behaviorism.
Computer simulation models emphasize strictly sequential operations, using flow charts.
Simon announces that computers can ‘think’.
Symbol processing languages are introduced allowing some success at theorem proving, problem solving, etc.
Minsky and Pappert kill off Perceptrons.Cognitive psychologists distinguish between algorithm and hardware.Neisser deems physiology to be only of ‘peripheral interest’Psychologists investigate mental processes as sequences of discrete stages.Slide4Slide5
Ubiquity of the Constraint SatisfactionProblem
In sentence processing
I saw the grand canyon flying to New York
I saw the sheep grazing in the field
In comprehension
Margie was sitting on the front steps when she heard the familiar jingle of the “Good Humor” truck. She remembered her birthday money and ran into the house.In reaching, grasping, typing…Slide6Slide7
Graded and variable nature of neuronal responsesSlide8
Lateral Inhibition in Eye of Limulus (Horseshoe Crab)Slide9
The Interactive Activation ModelSlide10
Interactive activation and probabilistic computationRumelhart’s
first effort to understand context effects in perception was formulated in explicitly probabilistic models.
Although he abandoned this formulation in favor of a neural network formulation, we will see that neural network models very similar to the IA model can be understood in explicit probabilistic terms.
Likewise, neural network models can be related to other sorts of models, including the drift-diffusion model of decision making and exemplar models of categorization and memory.
One of the goals of the course this year will be to explore these linkages more fully.Slide11
Synaptic Transmission and LearningLearning may occur by changing the strengths of connections
.
Addition and deletion of synapses, as well as larger changes in dendritic and axonal arbors, also occur in response to experience
.
[Recent
evidence suggests that neurons may be added under certain circumstances.]
Pre
PostSlide12
Connection-based learning creates implicit knowledge
Connection adjustment affects processing, not necessarily conscious awareness.
But not all learning is implicit.
Connection based learning can also be used to reinstate patterns of activation or to ‘auto-associate’ some elements of a pattern with other elements.
Perhaps we are aware of the patterns, but not of the connections that support their activation.Slide13
Cognitive Neuropsychology (1970’s)Geshwind’s disconnection syndromes:
Conduction Aphasia
Patient can understand and produce spoken language but cannot repeat sentences or nonwords
Alexia without Agraphia
Deep and surface dyslexia (1970’s):
Deep dyslexics can’t read non-words (e.g. VINT), make semantic errors in reading words (PEACH -> ‘apricot’)Surface dyslexics can read non-words, and regular words (e.g. MINT) but often regularize exceptions (PINT).Work leads to ‘box-and-arrow’ models, reminiscent of flow-chartsSlide14Slide15
Graceful Degradation in Neuropsychology
Patient deficits are seldom all or none
This is true both at the
task
and at the
item level.Performance is slower, more errorful, and requires more contextual support.And error patterns are far from random:Visual and semantic errors in deep dyslexia suggest degradation, rather than loss of a module or disconnectionRegularization errors depend on a word’s frequency, and how many other exceptions there are that are like it
Effects of lesions to units and connections in distributed connectionist models nicely capture these features of neuropsychological deficits.Slide16
Core Principles of Parallel Distributed Processing
Processing occurs via interactions among neuron-like processing units via weighted connections.
A representation is a pattern of activation.
The knowledge is in the connections.
Learning occurs through gradual connection adjustment, driven by experience.
Learning affects both representation and processing.
H I N T
/h/ /i/ /n/ /t/Slide17
Implications of this approach
Knowledge that is otherwise represented in explicit form is inherently implicit in PDP:
Rules
Propositions
Lexical entries…
None of these things are represented as such in a connectionist/PDP models.Knowledge that others have claimed must be innate and pre-specified domain-by-domain often turns out to be learnable within the PDP approach.
Thus the approach provides an alternative to other ways of looking at many aspects of knowledge-dependent cognition and development.
While the approach allows for structure (e.g. in the organization and interconnection of processing modules; structured similarity relations among patterns of activation), processing is generally far more distributed, representation is less explicit, and causal attribution becomes more complex.Slide18
In short…
Models that link human cognition to the underlying neural mechanisms of the brain simultaneously provide alternatives to earlier ways of understanding processing, learning, and representation at a cognitive level.Slide19
The PDP Approach…Attempts to explain human cognition as an emergent consequence of neural processes.
Global outcomes, local processes
Forms a natural bridge between cognitive science on the one hand and neuroscience on the other.
Is an ongoing process of exploration.
Depends critically on computational modeling and mathematical analysis.Slide20
Beyond PDP
Since the PDP work began, several new approaches and communities have arisen
NIPS/Machine Learning Community
Computational Neuroscience Community
Bayesian Approaches in Cognitive Science and Cognitive Neuroscience
Many of the models we consider belong more to these communities than to what might be called ‘Classic PDP’Much of my own work now involves eitherConstucting models at the interface between PDP and other approaches
Attempting to understand the relationship between PDP models and models formulated in other frameworks, including Bayesian approaches.
A good fraction of the course material will cover work of this type, and links between such work and PDP models.Slide21
This course…
Invites
you
to join the ongoing exploration of human cognition using PDP models and related approaches to mind, brain, and computation.
Focuses ultimately on human cognition and the underlying neural mechanisms, rather than abstract computational theory or artificial intelligence.
Includes exercises that provides an introduction to the modeling process and its mathematical foundations, preparing you to join the ongoing exploration.Slide22
Assignment for WednesdayRead:
* McClelland
, J. L. (2013).
Bayesian inference, generative models, and probabilistic computations in interactive neural networks.
Draft, Jan. 6. 2013, Department of Psychology, Stanford University.
Pages 1-28.and for a primer on real neurons:† Kolb, B. and Whishaw, I. Q. (1980). Physiological organization of the nervous system. Chapter 2 of Fundamentals of Human Neuropsychology (pp. 31-42). San Francisco: Freeman.We will discuss connectionist units and their properties in relation both to Bayesian computations and physiology of real neurons.