Overview of Cognitive Neuroscience

1
Overview of Cognitive Neuroscience

• Russell A. Poldrack
• UCLA Department of Psychology and Brain Research
  Institute

2
What is cognitive neuroscience?

• The science of how the brain creates the mind
• The questions of cognitive neuroscience are
• What are the neural building blocks of cognition?
• What is the computational function of these
  building blocks, and how do they relate to one
  another
• There are many possible levels of analysis in
  neuroscience
• We need to choose the right level for the
  question at hand

3
Levels of analysis in neuroscience
1 m
CNS
systems
10 cm
maps
1 cm
networks
1 mm
neurons
100 µm
synapses
1 µm
molecules
10 nm
After Churchland Sejnowski, 1992
4
How might we study the brain?

• What are the experimental methods of
  neuroscience?

5
How might we study the brain?

• Phrenology
• Studied brain function by measuring the size of
  bumps on the skull
• Early phrenologists were true scientists, but
  later it was adopted by charlatans

Source History of Phrenology on the Web
6
The fate of phrenology

• What it got right
• Mental functions can be localized to particular
  regions in the brain
• What it got wrong
• Method was subjective and qualitative
• Open to bias, both intentional and unintentional
• One basic assumption was incorrect
• The shape of the skull does not reflect the shape
  of the brains surface
• Another basic assumption was mostly wrong
• Size of region does not reflect the function of
  that region
• However, the size of cortical regions does
  reflect experience
• Basic psychology was incorrect
• Localization is not based on the kind of mental
  faculties identified by phrenologists (e.g.,
  philoprogenitiveness)

7
How might we study the brain?

• Lesion method
• Study of patients with brain lesions
• Due to stroke, tumor, injury
• Came to prominence in late 1800s with work of
  Broca, Wernicke, and others
• Also used to study brain function in experimental
  animals

8
Virtual lesions using TMS

• Transcranial magnetic stimulation
• Allows disruption or alteration of cortical
  function by induced electrical activity

9
Lesion method

• Strengths
• Allows determination of whether a particular
  brain region is necessary for a particular
  cognitive function
• Lesions can be verified (using imaging or
  post-mortem analysis)
• Can be used in animals to produce very
  anatomically precise lesions
• Weaknesses
• In humans, lesions are usually large messy
• Some regions are more susceptible than others
• Lesions usually affect underlying passing fibers
• Does the lesion effect reflect destruction or
  disconnection?

10
How might we study the brain?

• Neurophysiology
• Measurement of electrical activity
• In humans, implanted electrodes are used in
  neurosurgical planning
• Either implanted or on cortical surface during
  craniotomy
• In animals, arrays of electrodes can be implanted
  for extended periods for recording

Source Penfield Rasumussen (1952)
11
Neurophysiology

• Strengths
• Allows direct measurement of neural activity
• High temporal and spatial resolution
• Weaknesses
• Limited to a small number of neurons
• In humans, use is limited to patients who have
  brain pathology
• The region being investigated is often abnormal

12
How might we study the brain?

• Neuroimaging
• Noninvasive techniques for measurement of brain
  activity
• What aspects of brain function can be measured
  non-invasively?
• Electrical activity
• Electroencephalography (EEG)
• Magnetoencephalograhy (MEG)
• Blood flow
• Positron Emission Tomography (PET)
• Functional magnetic resonance imaging (fMRI)
• Optical imaging

13
Behavior (accuracy, RT)
14
The goal of cognitive neuroscience is to
establish these links - But how?
15
Forward inference
16
Reverse inference
17
Assessing reverse inference using the BrainMap
database

• Does activation in Brocas area imply that
  language is engaged?
• Searched for studies showing activation in
  Brocas area
• Look for all studies coded as Language studies

Poldrack, 2006, TICS
18
P(Brocas)0.113
P(BrocasLang)166/(166703)0.191
P(BrocasLang)199/(199 2154)0.085
P(
e.g. P(LangTaskJ)0.5 P(LangTaskJ,
Brocas)0.69
Bayes Factorposterior odds/prior odds2.3 (BFlt 4
considered weak)
Poldrack, 2006, TICS
19
Experimental designs

• Goal
• Determine how cognitive processes relate to brain
  systems
• How can we do this?
• Simple approach
• Subtraction
• More complex approaches

20
Simplest approach The subtraction method

• Acquire data under two conditions
• These conditions should differ only in the
  cognitive process(es) of interest
• Compare brain images acquired during those
  conditions
• Regions of difference reflect activation due to
  the process of interest

Petersen et al., 1988
21
The pure insertion assumption

• Subtraction requires a strong assumption of pure
  insertion
• Insertion of a single cognitive process does not
  affect any of the other processes
• Cf. Additive factors logic for reaction time
  (Donders/Sternberg)
• Failure of PI means that the results may not
  actually reflect the specific cognitive process
  being manipulated
• There are many examples of how this might fail

22
Failure of the PI assumption

• Jennings et al. (1997)
• Compared semantic and letter judgment tasks with
  three different response modalities (mouse,
  vocal, and silent mental)
• Task and response modality interacted in left
  prefrontal cortex

LIPC
23
More complex designs

• Parametric designs
• Vary a single parameter over a range of values
• Factorial design
• Combine several variables factorially and examine
  their interaction
• Repetition designs
• Examine the effects of repeating the same
  stimulus/task

24
Things to keep in mind

• Task design is critical
• Lack of proper control can lead to confounds that
  make interpretation of results difficult or
  impossible
• Behavior is critical
• Differences in task difficulty can lead to
  non-specific differences in activation

25
Overview of course

• Morning
• Basics of neurobiology, computation, and
  development
• Afternoon
• Review of specific systems