Cognitive Neuroscience Flashcards
how do we use the damaged brain to learn about the healthy brain?
we study people with brain damage in certain areas and compare them to people without damage in order to draw conclusions about what the damaged area of the brain does e.g. Phineas Guage
why should we study the brain?
studying the brain at a neural level allows us to understand cognitive processes and help when disorders arise
how do brain scans allow for communication?
EEG and fMRIs have allowed (some) communication with people in vegetative states as they can compare the ‘yes’ and ‘no’ brain activity to that of a healthy subject
how do we research cognitive function and neural activity?
to measure their relation to each other, we can either:
- change behaviour and see its effect on the brain
- or change the state of the brain and see its effect on behaviour.
recording methods
manipulating behaviour and measuring brain activity = correlation technique
inference methods
manipulating brain state (testing people with lesions) and measuring the effect on behaviour = causal technique
are inference or recording studies better?
inference studies allow stronger inference about necessity of a brain region but recording studies allow for greater flexibility in experimental design and are often a richer data source
why do we need different techniques of measuring the brain?
each technique is good for something but poor in something else - sometimes combining multiple brain measuring techniques yields the best result = converging evidence
classical neuropsychology
an approach that involves mapping brain areas to cognitive functions
- typically performed at a group level
- good at answering clinical questions
cognitive neuropsychology
an approach that involves determining whether functions dissociate under damage
- often relies heavily on case studies
- focuses on cognitive processes
- doesn’t require information on where the damage is
what is a single dissociation
we find a patient with a impairment in on aspect of cognition e.g. they can read easy made-up words but struggle difficult regular words.
this does not allow us to conclude anything as reading irregular words might just eb harder than non-words. we need a double dissociation to make conclusions
what is a double dissociation
if we want to rule out the difficulty of a task as a factor influencing a patients performance, we need to find another patient who displays the opposite deficit to the original.
for example, if patient X can read non-words but struggles with irregular words, we need to compare them to a patient Y who struggles with non-words but can read irregular words.
this would allow us to conclude that the systems that deal with irregular words and non-words are separate in the brain.
assumptions in single case research
fractional assumption = brain damage can selectively affect different cognitive/neural systems
transparency assumption = brain lesions can affect existing cognitive systems but do not create new systems
universality assumption = all cognitive systems are the same
dual route model of reading
- connectionist ‘triangle’ model suggests we read semantically (we have to activate the meaning of irregular words in order to read them)
dual-route cascaded model suggests we use lexical representations to read irregular words
neuropsychology could help solve this debate
neuropsychology and reading in people with dementia
100 observations in semantic dementia:
- Impaired semantic knowledge of word meanings
- Poor irregular word reading
Suggests that semantic knowledge is necessary to pronounce irregular words correctly
BUT what if patients have damage to separate semantic and lexical systems that are close to each other in the brain?
this is where single case studies come in (EM)
neuropsychology and reading - EM
EM could read irregular words but had impaired semantic knowledge - suggesting semantics are not necessary to read complex words
BUT is this single case study enough to disregard the association seen in 100 people with dementia?
Woollams et al. claimed it wasn’t but offered an alternative explanation: that there are individuals differences in how much people rely on the semantic route to read
How do neurons communicate?
they receive electric potentials (excitatory or inhibitory) from other neurons and, once a certain threshold is surpassed, an action potential is passed along the axon which triggers neurotransmitters at the synapses to communicate with other neurons
measuring neural action in animals
often with animals we are able to measure the firing rates of individual neurons with single cell recordings
measuring neural action in humans
it is more difficult to do single cell measures in humans so instead we measure the summed activity of large populations of neurons with EEG and MEG
single cell recordings in animals
electrodes are surgically implanted into the brains of experimental animals - typically rodents or primates - and the firing rate is measured in response to stimulus
feline visual system (Hubel and Wiesel, 1950s-60S)
initially discovered by accident that neurons in the visual cortex are tuned to a particular orientation which gave rise to the idea that simple cells put information together to create a bigger picture
selectivity and hierarchical organisation in neurons
- selectivity = specific neurons respond to particular types of visual stimulus
- hierarchical organisation = higher-level neurons respond to increasingly complex stimuli
this is what lead to the grandmother cell hypothesis
face selective neurons (Chang and Tsao, 2017)
Electrodes implanted into inferotemporal (IT) cortex
Spiking represented by auditory clicks
e.g. Jennifer Aniston cell (Quiroga et al, 2005)
is there really a grandmother cell?
there are neurons that have been shown to fire to specific faces but its hard to say whether they’re truly specific for a single individual.
a neuron for each person you know seems implausible
Electroencephalography - EEG
place electrodes on the scalp and record changes in electrical potentials - these potentials are tiny (2-10 microvolts) so it takes a large number of neurons firing at the same time to get a response
- EEG has poor spatial resolution (this is what fMRI is for) as the electrode detecting the signal may be distant from the source
- EEG is more useful for learning about when neural activity occurs rather than where
- EEG signals are noisy e.g. random neuron firing/activity from eye movement or response to surroundings
Event related potentials - ERPs
Different types of stimuli tend to produce characteristic ERPs at different points in time
These are generally given names that reflect their polarity and timing in ms (N170, P600, N400 etc.)
polarity depends on spatial arrangement of neurons generating the activity
e.g. N170 means there’s a negative wave occurring at 170ms
ERPs are often used to track the time course of cognitive processes involved in a task.
e.g. face processing requires perception, recognition and identification.
perceptual coding of the face
N170 (face detector) is larger for human and animal faces compared with other objects (Rousselet et al, 2004)
facial recognition
N250 is larger for familiar compared with unfamiliar faces (Herzmann et al., 2004)
Magnetoencephalography - MEG
Same basic principle of EEG but signals are measured by SQUID (superconducting quantum interference device) sensors that record fluctuations in the magnetic field.
EEG = electrical field, MEG = magnetic field
- has better spatial resolution than EEG
BUT - not widely available and £££
- easily disrupted by ‘noise’