Transcript of The Opposite of Attention is Epilepsy: Six ways of thinking about attention and why you should.
- Slide 1
- The Opposite of Attention is Epilepsy: Six ways of thinking
about attention and why you should
- Slide 2
- Everyone Knows What Attention Is Everyone knows what attention
is. It is the taking possession by the mind in clear and vivid
form, of one out of what seem several simultaneously possible
objects or trains of thought...It implies withdrawal from some
things in order to deal effectively with others William James
- Slide 3
- Everyone Knows What Attention Is James definition is a good
start toward operationalizing attention What we would call
selective attention As distinct from arousal alerting cognitive
effort
- Slide 4
- Everyone Knows What Attention Is The goal of this talk: Give
you some conceptual handholds to start thinking about attention
Tell you my (and my labs) line of thinking about attention
- Slide 5
- Everyone Knows What Attention Is Preview: 1.Attention solves
the philosophical problem of a unitary consciousness 2.Attention
solves a metabolic and thermal engineering problem 3.Attention
solves a signal-to-noise problem by filtering noise 4.Attention
solves a signal-to-noise problem by boosting signal gain
5.Attention solves a problem of ambiguity 6.Attention solves a
network complexity problem
- Slide 6
- Everyone Knows What Attention Is Notice that none of these
ideas are exclusive of any others They are all compatible ways to
conceptualize the same phenomenon
- Slide 7
- Attention Solves a Philosophical Problem Unitary Consciousness
We are each one conscious self Attention is the phenomenological
manifestation of this constraint You only get one train of thought
Maybe attention is epiphenomenal If you are a neurophilosopher,
this is why you should think about attention.
- Slide 8
- Attention Solves an Engineering Problem Waste heat is both an
applied and theoretical characteristic of computation The human
brain is metabolically demanding The cortex is a thermal
engineering nightmare If you design microprocessors, this is why
you should think about attention.
- Slide 9
- Attention Solves a Signal-to-Noise Problem by Filtering Noise
capacity limit or sensory bottleneck notion first proposed by
Donald Broadbent in the 60s leaky filter notion proposed by Anne
Treismann in the 70s
- Slide 10
- Attention Solves a Signal-to-Noise Problem by Filtering Noise
Evidence: Selective attention acts as a gate to awareness Simons
& Levin If you do cognitive psychology (or neurophilosophy),
this is why you should think about attention.
- Slide 11
- Attention Solves a Signal-to-Noise Problem by Filtering Noise
What are the neural correlates of such gating?
- Slide 12
- Attention Solves a Signal-to-Noise Problem by Filtering Noise
Chelazzi et al. Evidence: Selective attention suppresses neurons
representing task- irrelevant features or objects Note that search
array always contains a good stimulus for the recorded cell but
that might not be the target
- Slide 13
- Intracranial Recordings of Attentional Selection Initial
response of cells is classical
- Slide 14
- Intracranial Recordings of Attentional Selection Initial
response of cells is classical Response during delay maintains a
representation of the target feature
- Slide 15
- Intracranial Recordings of Attentional Selection Initial
response of cells is classical Response during delay represents the
target feature Initial response to search array is classical
- Slide 16
- Intracranial Recordings of Attentional Selection About 200 ms
after array onset, response of cell begins to depend on attention
Response becomes more vigorous if cell is tuned to features of the
target (i.e. the selected stimulus) Response becomes suppressed if
cell is tuned to a non-target distractor If you do
electrophysiology: This is why you should think about attention!
(note this effect is absent in anesthetized animals)
- Slide 17
- Attention Solves a Signal-to-Noise Problem by Boosting Signal
Gain Evidence: responses are faster and more accurate (memory !)
for attended relative to unattended events If youre a cognitive
psychologist, this is why you should think about attention (and
probably you already do).
- Slide 18
- Attention Solves a Signal-to-Noise Problem by Boosting Signal
Gain Evidence: Event-Related Potentials are enhanced for attended
relative to unattended stimuli If you do electrophysiology, this is
why you should think about attention.
- Slide 19
- Attention Solves an Ambiguity Problem Sensory Input Ambiguity
Cell tuned to red. Should it fire? Area V4 Receptive field = ~4 deg
visual angle
- Slide 20
- Attention Solves an Ambiguity Problem Sensory Input Ambiguity
Cell tuned to red. Should it fire? Area V4 Receptive field = ~4 deg
visual angle If you do computational neuroscience, This is why you
should think about attention.
- Slide 21
- Attention Solves an Ambiguity Problem Response Mapping
Ambiguity (e.g. Stroop Task) Cell tuned to line orientation. Should
it affect your response? Area V4 Receptive field = ~4 deg visual
angle If you do computational neuroscience, This is why you should
think about attention. B L U E
- Slide 22
- Attention Solves an Ambiguity Problem Reward mapping ambiguity
Cell tuned to red. Should it be associated with reward? Area V4
Receptive field = ~4 deg visual angle If you do computational
neuroscience, This is why you should think about attention.
- Slide 23
- Attention Solves a Network Complexity Problem The brain is a
massively interconnected network - each neuron makes ~ 1000
connections Gordon Kindlmann & Andrew Alexander University of
Wisconsin Van Essen, Andersen & Felleman (1992)
- Slide 24
- Attention Solves a Network Complexity Problem On the time scale
of behaviour, the network is anatomically hard-wired Fast
functional reconfiguration
- Slide 25
- Attention Solves a Network Complexity Problem Point to the red
horizontal line
- Slide 26
- Attention Solves a Network Complexity Problem Point to the red
horizontal line Visual stimulus drives visual neurons Black Brain
Box Motor plan is executed
- Slide 27
- Attention Solves a Network Complexity Problem Point to the red
horizontal line Visual stimulus drives visual neurons Black Brain
Box Motor plan is executed
- Slide 28
- Attention Solves a Network Complexity Problem Point to the red
horizontal line Notice the mapping is selective:
- Slide 29
- Attention Solves a Network Complexity Problem Point to the red
horizontal line Notice the mapping is selective:
- Slide 30
- Attention Solves a Network Complexity Problem Now point to the
green vertical line Notice the mapping is easily reconfigured
- Slide 31
- Attention Solves a Network Complexity Problem
- Slide 32
- Thus sensory neurons are in some sense omnipotent each ones
contribution to cognitive and motor networks is not determined by
anatomical connectivity it is determined dynamically by some
control system
- Slide 33
- Attention Solves a Network Complexity Problem Notice this is an
extension of the binding problem Cells representing features of the
same objects must contribute to a reconstituted whole object
representation These cells must be bound to all the other cells
mediating the current cognitive or motor behaviour If you study the
connectome, this is why you should think about attention.
- Slide 34
- Attention Solves a Network Complexity Problem The brain is a
massively interconnected network - each neuron makes ~ 1000
connections
- Slide 35
- Attention Solves a Network Complexity Problem The brain is a
massively interconnected network - each neuron makes ~ 1000
connections X 1000
- Slide 36
- Attention Solves a Network Complexity Problem The brain is a
massively interconnected network - each neuron makes ~ 1000
connections X 1000
- Slide 37
- Attention Solves a Network Complexity Problem The brain is a
massively interconnected network - each neuron makes ~ 1000
connections X 1000
- Slide 38
- Attention Solves a Network Complexity Problem Crude Analogy By
4 synapses the tree comprises more than 10 Billion cells! Attention
prevents runaway connectivity: Clearly the brain must have a system
by which information is routed appropriately through the
network
- Slide 39
- Attention Solves a Network Complexity Problem What does runaway
connectivity look like? Heres a hint: the feed forward sweep of
signal following a visual event is relatively unconstrained by
attention Red = earliest response at this latency Yellow = has
already responded Lamme (2000) By ~115 ms post-stimulus, much of
the cortex has responded to the visual event
- Slide 40
- Attention Solves a Network Complexity Problem What would be the
consequence if attention did not select cell assemblies? Neural
Gridlock? Maybe not the right concept.
- Slide 41
- Attention Solves a Network Complexity Problem The brain is a
system of coupled oscillators Driving such systems can trigger
unexpected synchronization
- Slide 42
- Attention Solves a Network Complexity Problem Classic Example
of spontaneous synchronization
- Slide 43
- Attention Solves a Network Complexity Problem See a fabulous
TED talk about synchronization by Steven Strogatz at:
www.ted.com/talks/steven_strogatz_on_sync.html
- Slide 44
- Attention Solves a Network Complexity Problem Do brains exhibit
runaway global synchronization? Yes, this is characteristic of
certain kinds of epileptic seizures. 3 Hz Spike and Wave EEG
pattern during absence seizure
- Slide 45
- Attention Solves a Network Complexity Problem OK so how might a
brain solve this problem? How might the attention system facilitate
a dominant cell assembly and suppress others? Neuronal
communication through neuronal coherence - Pascal Fries, TINS
(2005)
- Slide 46
- Attention Solves a Network Complexity Problem Individual
oscillators coupled to a central oscillator
- Slide 47
- Attention Solves a Network Complexity Problem Role of the
central oscillator has been called the dominant network
Communication-through-coherence suggests that oscillations within
cell assemblies become phase locked One set of such assemblies
achieves global dominance by having their individual phases nudged
into coherence
- Slide 48
- Thanks To my lab past, present and future: Greg Christie Andrew
Butcher Jarrod Dowdall Karla Ponjavic Scott Oberg Dillon Hambrook
Amanda McMullen Sheena McInnes Aja Mason