ELECTROPHYSIOLOGY Flashcards
How is the world out there perceived, comprehended and acted by neurons in the brain?
*Neuroscience
Neural representation
The way in which objects and properties of the outside world manifest themselves in the neural signal (e.g. different spiking rates for different stimuli)
How is the world out there perceived, comprehended and acted by neurons in the brain?
**Cognitive psychology
Mental representations
The way in which objects and properties of the outside world manifest themselves in the mind, Example: the image of a person (grandmother).
Neuronal recording
(single and multi unit recording)
Electroencephalography
(changes in electrical potential on the scalp)
HOW DOES IT WORK? Single Cell and Multi-Unit Recordings
Invasive: it requires implanting a microelectrode in the brain tissue (normally conducted in animals).
intercellular vs extracellular and measurment of action potential
When neurons fire an action potential, the change in voltage is recorded by the electrode.
We can then relate neuronal activity to experimental events (stimuli/responses). By studying the relationship between neuronal responses and experimental events, we can investigate what cognitive processes the brain area we’re recording from is involved.
fine neuroanatomy - What can we measure?
Activity of a single neuron (single-unit recording)
Activity of multiple neurons (multi-unit recording)
Activity of a single neuron (single-unit recording)
We can look at firing rate: count of action potentials (“spikes”) per sec
Activity of multiple neurons (multi-unit recording)
We can look at the pattern of activity across neurons (e.g. synchronous firing)
What coding schemes does the brain use (Rolls & Deco, 2002)?
1 - Local representation
2- Sparse distributed
3- Fully distributed representation
Local representation
What coding schemes does the brain use
All the information about a stimulus is carried by one neuron (grandmother cell).
Sparse Distributed representation
What coding schemes does the brain use
All the information is carried by a few neurons in a population.
Fully distributed representation
What coding schemes does the brain use
All the information is carried by all the neurons in a population.
Grandmother cells
Single-cell recording in an epileptic patient during surgery. This neuron fires when the
patient sees a picture of a specific person (Halle Berry) but not pictures of other persons.
Electroencephalography (EEG)
Continuous recording of electrical activity in the brain
Event-related potentials (ERPs)
Non-invasive: main electrophysiological technique in humans
Electrodes placed on the scalp
Measures summed electrical potentials
from millions of neurons
Note: electrical signals from a single neuron are too small to record non-invasively and can’t be distinguished from signals from other neurons.
Event-related potentials (ERPs)
?
How/Why can we detect electrical
activity of neurons on the scalp?
Synchronously active populations of neurons generate a large enough electrical field
Not measuring action potentials; instead postsynaptic potentials in dendrites
the signal from a single neuron is too small to detect at the scalp
So how do we detect signals at all?
Synchronously active neurons: spiking at the same time
Physiological origin of EEG signal is the postsynaptic potentials in dendrites of cortical neurons (not action potentials)
Electroencephalography (EEG)
EEG electrode arrangement
The 10–20 system of electrodes used in a typical EEG/ERP experiment.
Predictable EEG for different behavioral states
Important in clinical investigations as an indicator of:
Alertness
Brain Health (e.g. epilepsy)
*NOT widely used in cognitive neuroscience.
USING EEG TO STUDY COGNITION
In a continuous EEG recording, can’t tell what part of the signal has to do with the cognitive process of interest
What we can do is have the participant perform a task and look at the portions of the signal at critical periods (during stimuli or responses)
Say we’re interested in the brain’s response to hearing tones
What might we do if we’re interested in the brain’s response to the tones?
Kind of obvious: just look at the portions of the signal when person is hearing ones
Using EEG to study cognition: Event-related potentials (ERPs)
Activity from several trials is averaged together to extract signal (e.g. electrical activity associated with stimulus onset) from noise (e.g. background electrical activity)
Data for any one trial consists of response to stimulus + random noise
What happens when you average?
increase signal-to-noise ratio
?
Signal to noise ratio: like in math, a fraction
Here when noise decreases, SNR goes up.
How else could we make SNR go up?
Why does averaging increase SNR?
Assumptions
Noise is random, therefore, when averaged across trials
it should cancel out
Signal is not random and it
is similar across trials, so it
should not cancel out.
Noise is random: sometimes positive, sometimes negative
One moment you may be thinking about the weather, another thinking about what you had for dinner last night.
Average of zero
Like if I got you guys to all pick a random number between plus 10 and minus 10, the average would be zero
But signal is not zero – similar every time
So by averaging we’re reducing the noise, just like on the last slide, which increases the signal to noise ratio.
ERP COMPONENTS
Electrodes record a series of positive (P) and negative (N) peaks
Peaks referred to consecutively (P1, P2) or based on timing (e.g. P300 is positive peak around 300ms)
Features of interest in ERP
Amplitude: size of electrical change (microvolts, mV)
Number of neurons working together
Latency: time from onset of stimulus
How quickly neurons react
EXAMPLE OF FEATURES OF INFERENCE
NoGo N2: successful no-go trials
Neural activity associated with inhibiting habitual responses
N2 latency decreases and amplitude increases with age
Neurons associated with inhibition become more efficient with age
Advantages and Disadvantages of ERP
ERP signal is directly related to neural activity
I.e. it is a direct measure of neural activity
Electrical activity is conducted instantaneously to the scalp
ERP has an excellent temporal resolution
Can’t tell where the ERP signal is coming from
ERP has poor spatial resolution
Spatial Resolution of ERPs
Two problems:
Electrical signal from a particular source is distorted
When it hits the skull, tends to spread underneath the scull
Spatial Resolution of ERPs
The inverse problem
Given an electrical signal recorded at the scalp, can’t know where it’s coming from
Where in the cortex are the signals coming from? How many sources are there?
DIPOLES
Getting around the inverse problem (badly)
Some researchers use dipole modeling: this makes assumptions about the number of dipoles (regions of electrical activity aka brain areas)
Better still, if interested in spatial resolution, use another technique (e.g. fMRI or Magnetoencelography)
How are ERP components related to cognitive processes?
Recording electrical signal at the scalp, (but really interested in cognitive processes (e.g. perception, attention, memory). ) - (()) ESBER
How are ERP components related to cognitive processes?
In ERP, different peaks may approximately reflect the functioning of different cognitive stages
Not a simple relationship between ERP peak and cognition, because each peak is a sum of different electrical activities
Imagine a simple cognitive task like the go no-go task. Might require 3 cognitive processes:
Perceive the stimulus, make a decision, execute a motor response.
We get people to do this task, observe an ERP wave like this.
What’s your guess about how these ERP components are relate to those cognitive processes?
So that would mean that the timing of this component is related to the timing of that cognitive process.
You’d be wrong!
It’s not the case that this wave that we measure definitely represents some particular cognitive process
We can’t say that Peak 1 represents perceiving the stimulus
Imagine 3 cognitive processes located in 3 different parts of the brain, say one for perceiving the stimulus, one for making a decision, one for making a motor response
When each one of those is active, going to produce electrical changes
If their activity overlaps, what we measure at the scalp is the sum of the electrical activity of all components that are active at the same time
Positive ones will make the signal go higher, negative ones will tend to make the signal lower
ERP Summary
Based on EEG (electroencephalography) recordings
synchronized to some aspect of the event (e.g. onset of a stimulus)
EEG signal is averaged over many events
reduce effects of random neural firing
Electrodes record a series of positive and negative peaks
Timing (latency) and amplitude of the peaks may be related to different aspects of the stimulus and the task
Single and multi-unit recordings
Invasive; excellent spatial and temporal resolution
Measure action potentials from individual neurons
ERP
ERP - Overall summary -
Non-invasive; excellent temporal resolution, lower spatial resolution than single unit
Measure summed postsynaptic potentials from millions of neurons