pdp: motivation, basic approach

44
PDP: Motivation, basic approach

Upload: joelle-gillespie

Post on 30-Dec-2015

32 views

Category:

Documents


1 download

DESCRIPTION

PDP: Motivation, basic approach. Cognitive psychology or “How the Mind Works”. Information processing. Mental representations. Transformations. Perception / sensation. Action. Key questions:. What are the mental representations? What are the transformations and how do they work? - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: PDP: Motivation, basic approach

PDP: Motivation, basic approach

Page 2: PDP: Motivation, basic approach

Cognitive psychology

or

“How the Mind Works”

Page 3: PDP: Motivation, basic approach

Information processing

Perception / sensation Action

Transformations Mental representations

Page 4: PDP: Motivation, basic approach

Key questions:

• What are the mental representations?• What are the transformations and how do they work?• Where do these come from?

Page 5: PDP: Motivation, basic approach

Answers circa 1979

• Mental representations are data structures (symbols bound together in relational structures)

• “Transformation” processes are rules, like the instructions in a computer program.

• The rules and starting representations are largely innate• Learning is a process of building up new structured

representations from primitives and deriving (and storing) new rules.

Page 6: PDP: Motivation, basic approach

Why were (are?) these ideas so appealing?

Page 7: PDP: Motivation, basic approach

EG modular, serial processing in word recognition…

Their application is highly intuitive, logical, maybe even obvious!

Page 8: PDP: Motivation, basic approach

Explain generalization / generativity, central to human cognition!

EG: Today I will frump the car, yesterday I….______________ the car.

EG: Colorless green ideas sleep furiously!

Page 9: PDP: Motivation, basic approach

Can build mechanistic, implemented models of behavior…

Page 10: PDP: Motivation, basic approach

Explains modular cognitive impairments!

Pure alexia: Word representations gone!

Prosopagnosia: Face reps gone!

Category-specific impairment: Animals gone!

Broca’s aphasia: Phonological word forms gone!

And so on…

Page 11: PDP: Motivation, basic approach

So why isn’t this a class on symbolic cognitive modeling?

Page 12: PDP: Motivation, basic approach

W O R K

Page 13: PDP: Motivation, basic approach

EG: Today I will meep the car, yesterday I….______________ the car.

Page 14: PDP: Motivation, basic approach

Learning and development…

Page 15: PDP: Motivation, basic approach

Cognitive impairments not so modular…

Pure alexia: Can’t see high spatial frequencies.

Prosopagnosia: Abnormal visual word recognition.

Category-specific impairment: Can still name half the animals.

Broca’s aphasia: Can still produce single concrete nouns, can’t do grammar.

And so on…

But also Had a lot of Other Problems…

Page 16: PDP: Motivation, basic approach

Some issues with modular, serial, symbolic approaches• Constraint satisfaction, context sensitivity• Quasi regularity• Learning and developmental change• Graceful degradation

Page 17: PDP: Motivation, basic approach

Rumelhart

• Brains are different than serial digital computers.• Brain:

• Neurons are like little information processors• They are slow and intrinsically noisy…• …but there are many of them and they behave in parallel, not in serial.• It turns out that noisy, parallel computing systems are naturally suited to

some of the kinds of behavioral tasks that challenged symbolic theories of the time.

Page 18: PDP: Motivation, basic approach

In other words…

• Rather than thinking of the mind as some kind of unconstrained “universal computer,” maybe we should pay attention to the kinds of computations a noisy, parallel, brain-like system can naturally conduct.

• Paying attention to the implementation might offer some clues about / constraints on theories about how the mind works.

• Added bonus: Such theories might offer a bridge between cognition and neuroscience!

Page 19: PDP: Motivation, basic approach

Golgi Stain

Cell body

Dendrites(receive signals)

Axon(transmits signals)

Axon Terminal

Page 20: PDP: Motivation, basic approach

-

-

--

--

+

+

+

+

+

++ +

++

+ + + +

- - - - -----

+

Page 21: PDP: Motivation, basic approach

--

- -

--

+

+

+

+

+

+

Page 22: PDP: Motivation, basic approach
Page 23: PDP: Motivation, basic approach
Page 24: PDP: Motivation, basic approach

Membrane potential at axon hillock

DepolarizedHyperpolarized

p(Fi

ring

a sp

ike)

Page 25: PDP: Motivation, basic approach

Input: Depends on activation of sendingneurons and efficacy of synapses

Output: Train of action potentials at a particular rate

Weight: Effect on downstream neuron

Page 26: PDP: Motivation, basic approach
Page 27: PDP: Motivation, basic approach

Six elements of connectionist models:

1. A set of units2. A weight matrix3. An input function4. A transfer function5. A model environment6. A learning rule

Page 28: PDP: Motivation, basic approach

Six elements of connectionist models:

1. A set of units• Each unit is like a population of cells with a similar

receptive field.• Think of all units in a model as a single vector, with

each unit corresponding to one element of the vector.

• At any point in time, each unit has an activation state analogous to the mean firing activity of the population of neurons.

• These activation states are stored in an activation vector, with each element corresponding to one unit.

Page 29: PDP: Motivation, basic approach

Input

Output

Hidden [1 0 .51 .52 .45]

Input

Output

Hidden

Bias

[1 0 .51 .52 .45 1]

Page 30: PDP: Motivation, basic approach

Six elements of connectionist models:

2. A weight matrix• Each unit sends and receives a weighted connection

to/from some other subset of units.• These weights are analogous to synapses: they are

the means by which one units transmits information about its activation state to another unit.

• Weights are stored in a weight matrix

Page 31: PDP: Motivation, basic approach

Input

Output

Hidden

Bias

[1 0 .51 .52 .45 1]

Sending

Receiving

Page 32: PDP: Motivation, basic approach

Six elements of connectionist models:

3. An input function• For any given receiving unit, there needs to be some

way of determining how to combine weights and sending activations to determine the unit’s net input

• This is almost always the dot product (ie weighted sum) of the sending activations and the weights.

Page 33: PDP: Motivation, basic approach

Input

Output

Hidden

Bias

[1 0 ?? ?? ?? 1]

-.2

-1.8

0.5 0.2

1.1 -2.2

.1 .2 .3 .4 .2

Sending

Receiving

Page 34: PDP: Motivation, basic approach

Six elements of connectionist models:

Page 35: PDP: Motivation, basic approach
Page 36: PDP: Motivation, basic approach

Input

Output

Hidden

Bias

[1 0 ?? ?? ?? 1]

-.2

-1.8

0.5 0.22

1.1 -2.2

.1 .2 .3 .4 .2

Sending

Receiving

Page 37: PDP: Motivation, basic approach

Six elements of connectionist models:

5. A model environment• All the models do is compute activation states over units, given

the preceding elements and some partial input.• The model environment specifies how events in the world are

encoded in unit activation states, typically across a subset of units.

• It consists of vectors that describe the input activations corresponding to different events, and sometimes the “target” activations that the network should generate for each input.

Page 38: PDP: Motivation, basic approach

In1 In2 Out

0 0 0

0 1 1

1 0 1

1 1 0

X-OR function

Input

Output

Hidden

Bias

Input1 Input2 Hidden1 Hidden2 OutputInput1Input2Hidden1Hidden2Outputbias

Input1 Input2 Hidden1 Hidden2 OutputInput1Input2Hidden1Hidden2Outputbias

Page 39: PDP: Motivation, basic approach

• Note that the model environment is always theoretically important!

• It amounts to a theoretical statement about the nature of the information available to the system from perception and action or prior cognitive processing.

• Many models sink or swim on the basis of their assumptions about the nature of inputs / outputs.

Page 40: PDP: Motivation, basic approach

Six elements of connectionist models:

6. A learning rule• Only specified for models that learn (obviously)• Specifies how the values stored in the weight matrix

should change as the network processes patterns• Many different varieties that we will see:

• Hebbian• Error-correcting (e.g. backpropagation)• Competitive / self-organizing• Reinforcement-based

Page 41: PDP: Motivation, basic approach

In1 In2 Out

0 0 0

0 1 1

1 0 1

1 1 0

X-OR function

Input

Output

Hidden

Bias

Input1 Input2 Hidden1 Hidden2 OutputInput1Input2Hidden1Hidden2Outputbias

Input1 Input2 Hidden1 Hidden2 OutputInput1Input2Hidden1Hidden2Outputbias

Page 42: PDP: Motivation, basic approach
Page 43: PDP: Motivation, basic approach

Central challenge

• Given just these elements, can we build little systems—models—that help us to understand human cognitive abilities, and their neural underpinnings?

Page 44: PDP: Motivation, basic approach

One early example…