Week 3 - Methods of cognitive neuroscience

Reaction times

Time course of info processing, e.g. Posner’s letter-matching task, Sternberg efficiency of recognition memory, Reicher word superiority effect

Studying brain lesions

Lesions to a region should impair performance on tasks that assess that function, e.g. Adolphs study of patient SM who had Urbach-Wiethe disease and functionally didn’t have an amygdala

Computerized tomography (CT)

Computer reconstructs a 3D brain from a series of 2D images

Magnetic resonance imaging (MRI)

• Powerful magnetic field (0.5-3T)

• Constructs images of brain based on position of magnetically charged protons of hydrogen atoms in the blood

• Better spatial resolution than CT

Diffusion tensor imaging (DTI)

• Variation of MRI

• Used to study anatomic structure of axons forming brain’s white matter tracts - structural connectivity

• Measures density and flow of water molecules in brain

Temporal resolution

The accuracy to which you can measure when an event is occurring

• Brain damage is permanent so no temporal resolution

• EEG/MEF and single cell recording have ms resolution, so best candidates

• PET and fMRI have minutes and seconds resolutions, respectively

Spatial resolution

Refers to the accuracy with which you can measure where an event is occurring in the brain

• Lesions and functional imaging methods have resolution at mm level

• Single cell recordings have resolution at level of neuron

Invasiveness

Refers to whether equipment is located inside or outside of body

• PET is invasive because it requires injection of radio-labelled isotope

• Single cell recordings highly invasive

• TMS not invasive because coil located outside of body

Deep brain stimulation

Surgically implants electrodes into specific brain regions

• Can stimulate deep in subcortex

• Good for neuromodulation over extended periods of time

• Good for medication resistance Parkinson’s

Pharmacology

• Can give ppts agonist drug that simulates or mimics a neurotransmitter

• Or antagonist drug that blocks or reduces a neurotransmitter

• Low spatial resolution

Knockout procedure

Where animals are bred so that a gene or set of genes do or do not express themselves

Optogenetics

• Injection with a virus that creates light sensitive ion channels on specific neurons/receptors

• Can manipulate neural firing with wavelengths of light as ion channels open and close in response to specific wavelengths of light

Transcranial magnetic stimulation (TMS)

• Targeted magnetic pulses temporarily excite sugar-cube sized groups of neurons

• Allows increases or decreases in neural excitability and causal inference

• Creates a virtual lesion

• Can feel very itchy

Transcranial direct current stimulation (tDCS)

• Weak electrical current travels between two scalp electrodes influencing cortical excitability

• Not as irritating as TMS, but less spatial resolution than TMS

Single cell recordings

Measures the electrophysiological responses of a single neuron using a microelectrode system

• High spatial and temporal resolution

• But highly invasive

Electrocorticography (ECoG)

Ppts have brain surgery for another tissue and electrodes are surgically implanted on the surface of brain

• Great temporal and spatial resolution

• But highly invasive and not good generalisability

Electroencaphalography (EEG)

Measure scalp recorded electrical activity

• Tissues between neurons and scalp conduct electrical signals

• Good temporal resolution

• Spatial resolution not so good

Event-related potentials (ERPs)

Consistent patterns of EEG that are triggered by a stimulus

Magnetoencephalography (MEG)

Measures electrical activity produced by magnetic fields of neural activity

• Temporal resolution as good as EEG but better spatial resolution

• Expensive - requires magnetically shielded room and super cold temperatures

Positron emission tomography (PET)

Measures blood flow, oxygen use, and glucose metabolism

• Injection of radio-labelled oxygen or glucose into bloodstream

• Measures location where this material accumulates

• Good spatial resolution (1cm) but poor temporal resolution (1 min at best)

Functional magnetic resonance imaging (fMRI)

Studies brain function, where in brain activity happens

• Adapts MRI to register the magnetic properties of oxygenated and deoxygenated hemoglobin, allowing real-time blood flow to be imaged

• Higher spatial resolution than EEG/MEG but lower temporal resolution as takes a while for oxygen to be consumed

How MRI works

• Proton in hydrogen in constant motion, creating a tiny magnetic field

• MRI creates a powerful magnetic field and protons become oriented parallel to this

• Radio waves pass through magnetised regions and protons absorb energy from these waves

• This changes their orientation in a predictable direction

Multi voxel pattern analysis (MVPA) of fMRI data

Identifies distributed neural activity patterns across multiple voxels to distinguish between experimental conditions

Converging methods

Combining EEG and MRI can give insights into temporal and spatially precise neural activity

Connectivity

Correlation patterns of brain activity during rest or task

• There is not 1:1 connection between brain and cognition

• Most cognitive functions are a product of simultaneous activity of multiple brain regions

Single dissociation

• Damage to x impairs A, not B

• Could reflect task difficulty differences

Double dissociation

• Damage to x impairs A, not B

• But damage to y impairs B, not A

• Evidence that tasks rely on different neural systems

• Strongest neuropsychological evidence

Connectivity maps

• Also referred to as connectomes

• Visualisation of structural or functional connections within brain

• Computational and statistical problems as the whole brain fMRI contains 200,000 voxels, and this would require 40 billion correlations for each time point

Correlational methods

• fMRI

• PET

• EEG

• MEG

Causal methods

• Lesions

• TMS

• Pharmacological manipulation

• Optogenetics

Simulations

Reproduction of behaviour in alternative medium, designed to mimic behaviour and the cognitive processes supporting this

Neural networks

Info processing distributed over units whose inputs and outputs represent specific features

Deep learning models

Networks of many layers, representational diversity and complexity of these allow networks to recognise speech and faces