§ The electrophysiological brain 1
§ PSY2010
§ Representing the world
§ How is the world ‘out there’ perceived, comprehended and acted upon by a bunch of neurons ‘in here’?
§ Really is the main question neuroscience wants to answer
§ NB distinction:
ú Neural representation
ú Cognitive representation
ú (representation necessary because we don’t have entire world reproduced in our head)
§ Representing the world …
§ Neural representation – we will be talking about that in these classes; how neurons respond to stimuli, how they behave
§ Cognitive representation – mental representation of something (girlfriend, puppies, mathematics) within information processing model
§ These 2 kinds of representations unlikely to have 1- 1 relationship with each other
§ Representing the world …
§ Neuron’s rate of responding
§ i.e., number of action potentials or ‘spikes’
§ reflects something about what that neuron is doing with information
§ Single cell recordings
§ Possibly most important discovery in neurophysiology
§ Used to investigate questions related to neural representations
§ Measure response rate of neuron to stimulus
Primary goal of single-cell recording:
To determine experimental manipulations that produce a consistent change in the response rate of an isolated cell (increase or decrease)
§ Single cell recordings …
§ INVASIVE METHOD
ú Measures action potential
ú Normally only experimental animals (occasionally humans
during brain surgery)
during brain surgery)
ú Impossible to measure action potential from single neuron
non-invasively (i.e. from scalp) – signal too weak
non-invasively (i.e. from scalp) – signal too weak
§ By measuring changes in neuron’s responsiveness to changes in stimulus, can make inferences about how cognitive processing works
§ Single cell recordings …
§ Intracellular recording: electrode inside axon
§ Extracellular recording: electrode outside membrane
§ Multi-cell recording: can simultaneously record from up to 100 single neurons
§ Distributed vs sparse coding?
§ Hubel & Wiesel – showed that visual system is hierarchical
§ Starts from very simple (light and dark)
§ Builds from this (edges, lines)
§ Builds à more complex (shapes)
§ What is highest level of
hierarchy?
hierarchy?
§ One cell that responds to only
one thing?
one thing?
§ e.g. your grandmother?
§ Grandmother cells would store
all sorts of info abt that object;
not just visual info
all sorts of info abt that object;
not just visual info
§ 3 Possibilities for neural representation
1) Local representation
All the information about a stimulus/event = carried in one of neurons (i.e. grandmother cell)
2) Fully distributed representation
All the information about a stimulus/event = carried in all the neurons of a population
3) Sparse distributed representation
A small proportion of neurons carry information about a stimulus/event (i.e. Activity in several neurons required to represent a stimulus)
Rolls & Deco (2002)
§ Previous figures shows the responses of a single unit in the left posterior hippocampus to a selection of 30 out of the 87 pictures presented to the patient.
§ None of the other pictures elicited a statistically significant response. This unit fired to all pictures of the actress Jennifer Aniston alone, but not (or only very weakly) to other famous and non-famous faces, landmarks, animals or objects.
§ Interestingly, the unit did not respond to pictures of Jennifer Aniston together with the actor Brad Pitt. Notably, this unit was nearly silent during baseline and during the presentation of most other pictures.
§ Rate vs temporal coding
§ Rate coding: measures how neuron increases rate of firing in response to specific stimlulus
§ Temporal coding: looks to see whether neurons synchronize in their response
§ NB – Temporal coding may be one mechanism for integrating information across spatially separated populations of neurons
§ Evaluation
§ Information represented by neurons in terms of their rate of firing; and sometimes by synchronization
of firing
of firing
§ Sparse distributed coding of info is most likely – activity in several neurons required to represent
a stimulus
a stimulus
§ Sparse à hi memory capacity; low energy required
§ Distributed à protects against info loss (can lose a couple neurons/synapses and not lose info)
§ Distributed à allow cognitive system to generalize/categorize: novel stim that is similar to stored stim activates part of this representation)
§ The electrophysiological brain 2
§ PSY2010
§ Electroencephalography (EEG)
§ Neural activity = electrochemical process
§ EEG records electrical signals generated by brain through electrodes placed on
various points on scalp
(thus records overall brain activity)
various points on scalp
(thus records overall brain activity)
§ Non-invasive procedure and involves recording (not stimulation) – therefore harmless
§ How does EEG work?
§ Large populations of neurons active - produce electrical potentials that are large enough to be measured by electrodes on the surface of the scalp
§ EEG signal originates in post-synaptic dendritic currents (passive) rather than the axonal currents (active – Action Potential)
§ Electrodes = much larger than those used for single-cell measurements . Used with conductive gel
§ Measure change in voltage between signal at recording & reference electrode
§ Often arranged according to 10-20 system (Jasper, 1958)
§ Labelled according to location on scalp (i.e. F=frontal; P=parietal… C=central) and hemisphere involved (odd=left; even=right; z=midline)
§ How does EEG work? …
For electrical signal to be detected on scalp:
§ Whole population of neurons must be active in synchrony - generate large enough electrical field
§ Neurons must be aligned in parallel so that they summate rather than cancel out
§ Neuronal activity of deeper subcortical structures not easily measured
§ NB – the activity recorded at each scalp location cannot necessarily be attributed to neural activity near that region. Thus EEG not good for trying to localize activity
Inverse problem
§ Event-Related Potentials (ERPs)
§ Modern cognitive neuroscience aims to detect electrophysiological changes in response to particular stimuli/tasks (referred to as event-related potentials)
§ ERPs = based on EEG technique (but measures changes in electrical signal associated with specific event)
§ Useful for timing of events
§ Principle: when averaging different EEG waves over many presentations of a stimulus (i.e. 50-100 trials) – the signal-noise ratio is enhanced and an event-related potential is observed
§ Examining face processing with ERPs
Hypothetical stages of face-processing:
ú perceptual coding (N170)
ú facial identity computed (N250)
ú representation of identity of person (P400-600)
§ N170: - Perceptual Coding of face
ú not influenced by whether face is famous or not (also cartoons),
ú Reduced with perceptual degradation
ú N250: - Face recognition
ú Unaffected by view changes
ú affected by whether face is familiar/famous or not
ú Responds to diff images of same person
§ P400-600: - Person recognition
ú affected by both faces and names
§ Possible to use ERP markers to decide between
different theories
different theories
§ Associative/semantic priming: Reaction times = faster to stimulus if preceded by a stimulus of similar meaning
§ Debate in literature:
ú Does this effect arise from an early or late stage of processing?
ú Used ERP measures to determine the locus of associative priming of faces and names. Found associative priming has late effect - after 300ms – on ERP waveform
(more consistent with post-perceptual locus)
§ Mental chronometry
§ “chrono”= time; “metric” = measure
DEFINITION: The study of the time-course of information processing
in the human nervous system
in the human nervous system
Ø A basic assumption in cognitive psychology is that tasks are composed of a
set of mental operations
set of mental operations
Ø Changes in the efficiency of information processing → change in response time
Ø Not interested in absolute time, but rather relative differences in time
I.e:
ú 4 + 2 = 6 fast Reading House = faster
ú 4 + 3 = 7 slower OR vs
ú 4 + 4 = 8 slowest Reading HoUsE = slower
§ Conclude:
ú Mathematical sums not just stored as a set of facts
ú May need to transform new representation into more standard one
§ Saul Sternberg: Additive Factors method
§ Developed general method for dividing reaction times into different stages
§ His task:
ú WM task (hold 1-4 digits in mind) 2 7 4
ú Then show probe ® yes/no 7
ú DV = reaction time
§ Sternberg argued that task could be divided into 4 diff processing stages:
ú Encoding
ú Comparing
ú Decision
ú Responding
§ Evaluation
§ Investigating time-course of cognitive processes = NB Event-related potentials have excellent
temporal resolution
temporal resolution
§ Advantages over RT:
ú Provides continuous measurement of change over time rather than single timing measure
ú Enables electrophysiological changes associated with unattended stimuli to be measured
(RT always requires overt response)
(RT always requires overt response)
§ Magnetoencephalography (MEG)
§ Advantages:
§ - Same good temporal resolution as EEG
§ - Signal unaffected by skull, meninges (less influenced than electrical currents, EEG)
§ - In addition: can localize the source of the signal (spatial accuracy has made it useful in Neurosurgery; 2-3mm)
§ Limitations:
§ - Only able to detect electrical flow that is oriented parallel to surface of skull (only sulci)
§ - Very expensive & limited availability (uses liquid helium & magnetically isolated rooms)
§ The imaged brain
§ PSY2010
§ Recap of research methods covered so far:
§ Lesion studies
§ Reverse engineering
§ Examine what happens to cognitive function when part
of brain damaged
of brain damaged
§ Lesion studies and neuroimaging inform each other
§ Human vs animal work – human lesions not precise
§ TMS – virtual lesions in humans
§ Electrophysiological research methods (also = functional imaging)
§ Single cell recordings – where and when; mainly animals
§ EEG – general brainwave activity
§ ERP/mental chronometry – when
§ MEG – when and where
§ Summary of methods in neuroscience
§ Overview of 2 lectures on imaging:
A. How functional & structural brain imaging works
B. Methodological factors to ensure that results are meaningfully linked to cognitive theory
C. How functional imaging data is analysed to find regions of activation,
& possible pitfalls of interpretation
§ Structural imaging:
§ Construct static maps of brain
§ Based on fact that different types of tissue
(i.e. skull, grey matter, white matter etc)
have different physical properties
(i.e. skull, grey matter, white matter etc)
have different physical properties
§ Two most common methods:
§ CT
§ MRI
(Magnetic Resonance Imaging)
(Magnetic Resonance Imaging)
§ Functional Imaging:
§ Produce dynamic maps of moment-to-moment activity
of brain during cognitive processing
of brain during cognitive processing
§ Based on assumption that neural activity produces local physiological changes in relevant region of brain
§ Two most common methods:
§ PET
§ fMRI
§ The CT scan
§ Advanced version of conventional X-ray (can reconstruct 3D image)
§ Image constructed according to amount of x-ray absorption by different tissues.
Amount of absorption = related to tissue density:
ú bone absorbs most (skull appears white)
ú CSF absorbs the least (ventricles appear black)
ú brain matter = intermediate (appears grey)
§ Typically only used in clinical settings, i.e. to diagnose tumour or identify haemorrhage or other gross abnormalities
§ CSF = cerebrospinal fluid
§ Limitations:
ú Typical spatial resolution: 0.5-1cm
(Thus not possible to discriminate objects closer than about 5mm)
(Thus not possible to discriminate objects closer than about 5mm)
ú Difficult to distinguish between white & gray matter
ú Potentially dangerous: x-rays cumulatively absorbed by
high-density tissue
high-density tissue
ú Cannot be adapted for functional imaging purposes
§ Advantages:
ú Inexpensive
ú Obtain scan in relatively short period of time (+/- 3min)
ú Can add contrast agent – better detection of
blood-thirsty structures
blood-thirsty structures
§ MRI Physics
§ Protons in H-nuclei in constant motion, spinning about their principle axis ® magnetic dipole
§ MRI machine creates a powerful magnetic field (1.5 or 3 Tesla)
§ Protons become oriented in direction parallel to this
magnetic field
magnetic field
§ Brief radio frequency pulse flips the protons – causes them to align at 90 degrees to the main magnetic field of the scanner
§ After this pulse, protons ‘relax’ back to where they were before the RF pulse
§ This ‘precession’ of the protons from the flipped state to the parallel state gives the measurable signal that fMRI detects
§ This process repeated, RF pulse sent to excite different slices of the brain
§ Whole brain can be scanned in abt 2sec (3mm slices)
§ Different types of of MRI images
§ T1-weighted image: Based on variation in rate at which protons return to aligned state after RF pulse
§ T1 most frequently used for clear structural images (grey matter looks grey, white matter clearly distinguishable)
§ T2-weighted image: Based on decay of signal from interaction with nearby molecules
§ T2*-weighted image: Also based on decay of signal from interaction with nearby molecules, incorporates
deoxyhemoglobin distortions
deoxyhemoglobin distortions
§ T2* used for functional magnetic resonance images
Advantages of MRI over CT scan:
§ Non-invasive technique: no ionizing radiation
§ Much better spatial resolution (can discern gyri)
§ Better discrimination between white & grey matter
§ Can be used to detect functional changes (fMRI)
§ Physiology underpinning Functional Imaging
§ However: understand what you see!
§ The brain is always physiologically active
§ Important consequences for using physiological markers as basis of “neural activity”
§ Cannot simply view in scanner which areas are receiving blood
& using oxygen – this must happen for all neurons all the time
& using oxygen – this must happen for all neurons all the time
§ What does an “active” site mean?
§ It means greater activity relative to baseline
§ Select suitable baseline task
§ Positron Emission Tomography (PET)
§ How it works:
§ PET Advantages/Disadvantages:
§ Functional Magnetic Resonance Imaging (fMRI)
§ How it works:
§ Essentially the same as with traditional MRI
§ With fMRI: component of MRI signal that used here (T2*) is sensitive to magnetic properties of hemoglobin (carries O2 in bloodstream, when oxygen is absorbed, hemoglobin becomes deoxygenated)
§ Deoxyhemoglobin = more paramagnetic than oxygenated hemoglobin à distortions in signal ; these measured to work out concentration of deoxyhemoglobin in blood
§ The fMRI detectors measure the ratio of oxygenated to deoxygenated hemoglobin
§ This ratio is referred to as the BOLD effect:
ú blood oxygen level dependent contrast
§ The Hemodynamic Response Function
§ By continuously measuring the fMRI signal, it is possible to construct a map of changes in regional blood flow that are coupled to local neural activity:
In this eg see how different
regions of brain respond
to moving versus static
stimuli
In primary visual cortex,
no difference in response
BUT in MT
see response to movement
and not to static
In this eg see how different
regions of brain respond
to moving versus static
stimuli
In primary visual cortex,
no difference in response
BUT in MT
see response to movement
and not to static
§ Changes in BOLD signal are small – typically between 1 – 3% in 1.5T scanners
§ Temporal resolution fair, but limited by time HRF takes – several seconds
§ Cognitive processes can happen much quicker
§ Spatial resolution good – 1mm
§ Which Method is best?
Over last 10 yrs fMRI has largely taken over from PET in functional imaging experiments
§ Key advantages of fMRI:
ú better temporal & spatial resolution.
ú Event-related designs possible
ú Doesn’t use radioactivity
§ Disadvantages of fMRI:
ú Scanner = noisy (less attractive for auditory experiments)
ú Small movements can distort signal (avoid speech)
ú Some brain regions susceptible to signal distortion (areas near air voids) Therefore some regions difficult to image: OFC and some temporal regions
§ NB – see comparison between PET & fMRI on p56
No comments:
Post a Comment