Embed Size (px)
Citation preview
Learning in the Brain
1) Recognition and Representation in Cortex
Recognition Networks
QuickTime™ and aVideo decompressor
are needed to see this picture.
The Bottom-up view of visual processing.
Recognition Networks - DistributedRecognition Networks - Distributed
Face Blind! Bill's Face Blindness (Prosopagnosia) Pages - Introduction
Recognition Networks - ParallelRecognition Networks - Parallel
PET scans by Lawrence Parsons, Peter Fox, and Donald Hodges Universty of Texas, San Antonio
Left panel: the harmony condition activated the left side of the brain more than the right. It also activated inferior (or lower) regions of the temporal cortex as compared to the melody condition
Center panel: the melody condition activated both sides of the area called the temporal cortex (which is known to represent sound) to a much greater extent than did the rhythm and harmony conditions.
Right panel: much of the brain activation observed during the rhythm condition was in the cerebellum.
Recognition is Distributed: Music
The McGurk effect and parallel processing
Another version of McGurk
Yet another explanation
The McGurk effect and parallel processing
What about students with learning disabilities?
Integration of heard and seen speech: a factor in learning disabilitiesin childrenErin A. Hayesa,*, Kaisa Tiippanab, Trent G. Nicola, Mikko Samsb, Nina Krausa,c,d
Recognition Networks - HeterarchicalRecognition Networks - Heterarchical
What’s wrong with this diagram?
See Top to Bottom highlights
Illusions as top-down constraints on understanding images.
Shadow
Presidential Illusion
Illusions Reveal the Brain's Assumptions
Ball and Shadow
On Illusions From Top to Bottom
The idea of constraint satisfaction and pattern recognition
What does recognition require?
An integrated network, where meaning is constructed “heterarchically” –
from the interplay of sensory information, perceptual
constancies, and cognitive experience.
Neural Networks
• Processing is Distributed
• Processing is Distributed
• Processing is Distributed in Parallel
Neural Networks
• Processing is Distributed
• Processing is Distributed in Parallel
• Processing is Distributed Heterarchically
Neural Networks
What barriers do imagespotentially present for the learner’s
recognition systems?
SensoryPerceptualCognitive
What can we learn from the neural network idea?
Recognition Cortex
Recognition cortex (sensory –perceptual cortex in the posterior neocortex) is composed of neural networks adapted to recognize regularities in the environment.
Recognition cortex does not “respond” to these regularities, it generates them.
Recognition Cortex
Perception, memory and learning are all processes of categorization – none of them are possible without lumping things together which are in fact not the same.
Recognition Cortex
Perception, memory and learning are all processes of categorization – all of them require lumping things together which are in fact not the same.
Perception, memory and learning are all processes of generalization – all of them generalize from some things to others.
Concepts are merely categories and generalizations with names on them.
Structure in Recognition Cortex
Structure is sculpted by learning – specifically by learning the regularities of the environment
There is no optimal structure, the optimal structure is dependent on the kind of memory/perception/learning is needed.
Overall structure is a network of networks – there is specialization and localization determined by learning
There are many specialized networks – not 5 senses, but many. 30 visual networks alone.
What if structure were different?
• Parameters – what could we change, and what would be the effects of changes?
Post-Mortem on Neural Network
• Parameters – what could we change, and what would be the effects of changes?– Proportions – different relative numbers of layers
– Sensitivity – damping down or exciting up
– Location – different placement in larger network, different upstream and downstream layers
– Circuitry – interneurons, interconnections, top-down
Post-Mortem on Neural Network
• Parameters – what could we change, and what would be the effects of changes?– Proportions – different relative numbers of layers
– Sensitivity – damping down or exciting up
– Location – different placement in larger network, different upstream and downstream layers
– Circuitry – interneurons, interconnections, top-down
– Lesions!!!!
Post-Mortem on Neural NetworkQuestions
• What happens when we lesion?
Learning in Recognition Cortex
Learning is distributed.
There is specialization, localization.
Specialization is determined by complex interaction between biology and environment.
Learning in Recognition Cortex
Learning is parallel – learning happens all over the recognition cortex at the same time.
Many different things are learned, many different kinds of learning
Learning in Recognition Cortex
Because of parallel processing:
Complex is not necessarily faster than simple
Complex is not necessarily simpler than simple
Learning in Recognition Cortex
Learning is heterarchical.
Expectancies, top-down affect bottom-up.
Difficult to separate perception, memory, cognition.
Learning in Recognition Cortex
Neural Networks don’t fill up. The more a network has learned already, the more it can assimilate more. The “catch” more.
Neural Networks get more and more sculpted to the environment.
Learning in Recognition Cortex
Recognition cortex learns by changing its structure – by changing the connections between neurons that make up the network.
Learning is a change in the way the network acts.
Learning is a dynamic process not a static one.
Piaget was right – assimilation is accompanied by accommodation.
Learning in Recognition Cortex
Learn by examples, and counter examples. (spelling rules)
Neural Networks get more and more sculpted to the environment.
Neural Networks change proportions to map the environment.
Recognition Cortex
Post-Mortem on Neural Network
• http://mbeneuorblog.blogspot.com/
• http://thebrain.mcgill.ca/flash/index_a.html
From Posner and Raichle
Vygotsky in the Brain
Recognition Networks
Recognition is Distributed: Visual
Individual Differences
The Problem of Ruth:Individual Differences In
Recognition
The Problem of Dyslexia: Individual Differences in Recognition
• From Shaywitz et al
Principles of Universal Design
•Provide Multiple Means of Representation
Principles of Universal Design
•Provide Multiple Means of Representation
Provide sensory and perceptual options
Provide linguistic options
Provide cognitive options
What kind of Options?
1. Provide multiple examples
2. Highlight critical features
3. Highlight big ideas
4. Highlight critical relationships
Federal Style Building
Provide Multiple Examples:
Highlight Critial Features:
Provide Background Information
• low pitched roof • smooth facade • window openings with larger panes and louvered shutters • delicate columns and molding • exterior detail expressed only at an entrance • circular, oval, or octagonal room shapes • interior decoration such as garlands, swags, urns, and rosettes applied to walls • pastel colors
Background Information as a limit on recognition:
•
Strategic Networks
From Yarbus, 1967
Individual Differences in strategic processes
From Luria: Prefrontal Patient
The Problem of Ruth:Individual Differences In Strategic Learning and
Expression
Principles of Universal Design
•Provide Multiple Means of Representation
•Provide Multiple Means of Expression
Multiple Means of Expression
Provide options in the motor skills required for action and interaction
Provide options in the tools or media available for expression
Provide options in the scaffolds available during apprenticeships.
Universal Design for Learning
Learning new skills and strategies
Provide multiple models and mentors
Provide graduated supports and scaffolds that can be gradually “released”
Provide options in the contexts required for performance
Provide Ongoing Feedback
Affective Networks
1. Offer choice of content and tools
2. Provide adjustable levels of challenge
3. Offer choice of rewards
4. Offer choice of learning context
Affective Learning: Options
The Problem of Ruth:Individual Differences 3
The problem of a room full of Ruths:
Individual Differences Everywhere…
Principles of Universal Design
•Provide Multiple Means of Representation
•Provide Multiple Means of Expression
•Provide Multiple Means of Engagement
Recognition Learning
1. Provide multiple examples
2. Highlight critical features
3. Provide multiple media and formats
4. Support background knowledge
Affective Networks
The Problem of Ruth:Individual Differences 3
What part of the Brain do we read with?
The Tortoise and the Hare
This is the house that Jack built. This is the malt, that lay in the house that Jack built. This is the rat, that ate the malt, that lay in the house that Jack built. This is the cat, that killed the rat, that ate the malt, that lay in the house that Jack built. This is the dog, that worried the cat, that killed the at, that ate the malt, that lay in the house that Jack built. This is the cow with the crumpled horn, that tossed the dog, That worried the cat, that killed the rat, that ate the malt, that lay in the house that Jack built. This is the maiden all forlorn, that milked the cow with the crumpled horn, that tossed the dog, that worried the cat,that killed the rat, that ate the malt, that lay in the house that Jack built. This is the man all tattered and torn, that kissed the maiden all forlorn, that milked the cow with the crumpled horn, that tossed the dog, that worried the cat, that killed the rat, that ate the malt, that lay in the house that Jack built. This is the bishop all shaven and shorn,that married the man all tattered and torn, that kissed the maiden all forlorn, that milked the cow with the crumpled horn, that tossed the dog, that worried the cat, that killed the rat, that ate the malt, that lay in the house that Jack built. This is the cock that crowed in the morn, that waked the bishop, all shaven and shorn, that married the man, all tattered and torn, that kissed the maiden all forlorn, that milked the cow with the crumpled horn, that tossed the dog, that worried the cat, that killed the rat, that ate the malt, that lay in the house that Jack built. This is the farmer who sowed the corn, that kept the cock that crowed in the morn, that waked the bishop all shaven and shorn, that married the man all tattered and torn, that kissed the maiden all forlorn, that milked the cow with the crumpled horn, that tossed the dog, that worried the cat, that killed the rat, that ate the malt, that lay in the house that Jack built.
Word Reading in the Brain
What part of the Brain do we read with?
That depends ……
What part of the Brain do we read with?
That depends ……
What part of the Brain do we read with?
That depends
On the task
What part of the Brain do we read with?
That depends
On the task
And the learner.
FMRI Summary -Dyslexia
• From Shaywitz et al.
Principles of Universal Design
• Provide Multiple Means of Representation
– Within-modality
• Variation in size, loudness, color, speed, pitch, contrast, etc.
– Cross-modality
• Text-to-Speech
• Speech-to-text (Captioning), etc.
Principles of Universal Design
• Provide Multiple Means of Representation
• Provide Multiple Means of Expression
– Within-modality
• Expanded keyboard, sticky keys, single switch, etc.
– Cross-modality
• Voice Recognition, composition, etc.
Video watching in the Brain
2
3
4
Fixed versus
Malleable Media
Focal versus
Networked Media
The Concept of
Universal Design
www.cast.org
Speech Prism - [email protected]
eReader – www.cast.org
Bobby – www.cast.org/Bobby
Music – www.mp3.com
Learning in the Brain:
1) differs according to the task
What part of the Brain do we learn to read with?
The Tortoise and the Hare
This is the house that Jack built. This is the malt, that lay in the house that Jack built. This is the rat, that ate the malt, that lay in the house that Jack built. This is the cat, that killed the rat, that ate the malt, that lay in the house that Jack built. This is the dog, that worried the cat, that killed the at, that ate the malt, that lay in the house that Jack built. This is the cow with the crumpled horn, that tossed the dog, That worried the cat, that killed the rat, that ate the malt, that lay in the house that Jack built. This is the maiden all forlorn, that milked the cow with the crumpled horn, that tossed the dog, that worried the cat,that killed the rat, that ate the malt, that lay in the house that Jack built. This is the man all tattered and torn, that kissed the maiden all forlorn, that milked the cow with the crumpled horn, that tossed the dog, that worried the cat, that killed the rat, that ate the malt, that lay in the house that Jack built. This is the bishop all shaven and shorn,that married the man all tattered and torn, that kissed the maiden all forlorn, that milked the cow with the crumpled horn, that tossed the dog, that worried the cat, that killed the rat, that ate the malt, that lay in the house that Jack built. This is the cock that crowed in the morn, that waked the bishop, all shaven and shorn, that married the man, all tattered and torn, that kissed the maiden all forlorn, that milked the cow with the crumpled horn, that tossed the dog, that worried the cat, that killed the rat, that ate the malt, that lay in the house that Jack built. This is the farmer who sowed the corn, that kept the cock that crowed in the morn, that waked the bishop all shaven and shorn, that married the man all tattered and torn, that kissed the maiden all forlorn, that milked the cow with the crumpled horn, that tossed the dog, that worried the cat, that killed the rat, that ate the malt, that lay in the house that Jack built.
Learning in the Brain:
1) differs according to the task
2) differs according to prior learning
Learning in the Brain:
1) differs according to the task
2) differs according to prior learning
3) differs according to the individual
Learning to Recognize
Content Enhancements
Reduction in Degrees of Freedom
Marking Critical Features
Vygotsky, Part 2
Principles of Universal Design
•Provide Multiple Means of Representation or….Content Enhancement Routines
… routines are developed and used with the following explicit goals:
•meeting the needs of both the group and the individual
•maintaining the integrity of the content
•selecting the critical features of the content
•transforming the content in ways that promote student learning
•carrying out instruction in a partnership with students.
Learning Strategies
Models or Demonstrations
Reduction in Degrees of Freedom
Feedback
