Chapter 3. The electrophysiological brain

Representations
Properties of the world that are manifested in cognitive systems (mental representation) and neural systems (neural representation)

Single-cell  recordings
Measures the responsiveness of a neuron to a given stimulus, in terms of action potential per second. Its used to investigate questions related to neural representation.

Electroencephalography (EEG)
Measurements of electrical signals generated by the brain through electrodes placed on different points on the scalp. This method is particularly useful for measuring the relative timing of cognitive events and neural activity.

Event-related potential (ERP)
The average amount of change in voltage at the scalp that is linked to the timing of particular cognitive events.

Reaction time
The time taken between the onset of  a stimulus/event and the production of  a behavioural response

IN SEARCH OF NEURAL REPRESENTATIONS: SINGLE-CELL RECORDINGS

How are single-cell recordings obtained?

By measuring changes in the responsiveness of a neuron to changes in a stimulus or changes in a task, it is possible to make inferences about the building blocks of cognitive processing. Single-cell recordings can be obtained by implanting a very small electrode either into the neuron itself (intracellular recording) or outside the membrane (extracellular recording). This is an invasive method. As such, the procedure is normally conducted on experimental animals only. The method is occasionally conducted on humans undergoing brain surgery. 
Multi-cell recordings (multi-unit): the electrical activity (in terms of action potentials per second) of many individually recorded neurons recorded at one or more electrodes. Technology has now advanced such that it is possible to simultaneously record from 100 neurons in multi-electrode arrays.

Distributed vs. sparse coding

Grandmother cell: a  hypothetical neuron that just responds to one particular stimulus.
Rolls and Deco (2002) distinguish between three different types of representation that maybe be found at the neural level:

1. Local representation. All the information about a stimulus/event is carried in one of the neurons (as in a grandmother cell)
2. Fully distributed representation. All the information about a stimulus/ event is carried in a all the neurons of a given population.
3. Sparse distributed representation. A distributed representation in which a small proportion of the neurons carry information about a stimulus/event.

According to research the response is determined by how familiar a face is rather than its appearance, suggesting its is coding features that are relevant to memory. 

Rate coding: The informational content of a neuron may be related to the number of action potentials per second.

Temporal coding is a type of neural coding which relies on precise timing of action potentials or inter-spike intervals. Combined with traditional rate coding models, temporal coding can provide additional information with the same rate. Temporal coding may be one mechanism for integrating information across spatially separated population of neurons. 

ELECTROENCEPHALOGRAPHY AND EVENT-RELATED POTENTIALS

How does EEG work?
The physiological basis of the EEG signal originates in the postsynaptic dendritic currents rather than the axonals currents associated with the action potential. EEg records electrical signals generated by the brain through electrodes placed on different points on the scalp.
Basic requirement EEG:

1. A whole population of  neurons must be active in synchrony to generate a large enough electrical field
2. The population of neurons must be aligned in a parallel orientation so that they summate  together rather than cancel out. 
   (Orientation of neurons in  the thalamus renders its activity invisible to this recording method)

Analogous phenomena of rate/temporal coding can be found in the EEG data, although here they are measuring the summed electrical activity over millions of neurons.
Neurons that respond in synch are generally believed to be communicating with each other opposed to responding in isolation.

To gain an EEG measure one needs to compare the voltage between two or more different sites. A reference site is often chosen that is likely to be relatively uninfluenced by the variable under investigation. One common reference point is the mastoid bone behind the ears or a nasal reference; another alternative is to reference to the average of all electrodes. The electrodes are labeled according to their location (Frontal,  Parietal, Occipital, Temporal, Central) and the hemisphere involved (odd numbers for left and even numbers for right and "z" for the midline)

Event-related potentials (ERPs)

The most common use of EEG in cognitive neuroscience is the method known as ERP. The EEH waveform reflects neural activity from all parts of the brain. Some of this activity may specifically relate to the current task but most of  it will relate to spontaneous activity of other neurons that do not directly contribute to the task. As such, the signal to noise ratio in a single trial of EEG is very low. The ratio can be increased by averaging the EEG signal over many presentations of the stimulus, relative to the onset of a stimulus.  
Whether a peak is positive or negative (its polarity) has no real significance in cognitive terms. The polarity depends on the spatial arrangement of the neurons that are giving rise to the signal at that particular moment in time. 
Dipole: a pair of positive and negative electrical charges separated by a small distance. Dipoles from different neurons and different regions summate and conduct to the skull, and these give rise to the characteristic peaks and troughs of the ERP waveform. What is of interest in  the ERP waveform, in terms of  linking it to cognition, is the timing an amplitude of those peaks.

Rhythmic oscillations in the EEG signal

The EEG signal, when observed  over sufficiently long time scale, has a wave like structure. The EEG signal tends to oscillate at different rates that are named after letters of the Greek alphabet:

• Alpha: 7-14Hz
  
• Beta: 15-30Hz
• Gamma: 30+ Hz

MENTAL CHRONOMETRY IN ELECTROPHYSIOLOGY AND COGNITIVE PSYCHOLOGY

Mental chronometry can be defined as the study of the time-course of information processing in the human nervous system. The basic idea is that changes in the nature or efficiency of information processing will manifest themselves in the time it takes complete a task.

Additive factors method: a general method for dividing reaction times into different stages according to Sternberg. He proposed that the task could be divided into a number of separate stages, including:

1. Encoding
2. Comparing
3. Deciding
4. Responding

He further postulated that each of these stages could be independently influenced by different factors affecting the task.

Investigating face processing with ERPs and reaction times

As with the single-cell results, there is evidence for an ERP component that is relatively selective to the processing of faces compared with other classes of visual objects. This has been termed the N170 (a negative peak at 170ms) and is strongest over right posterior temporal electrode sites. This component is uninfluenced by whether the face is famous or not.
The N250, by contrast, is larger for famous and personally familiar faces relative to unfamiliar faces and responds to the presentation of different images of the same person.

One debate in the cognitive psychology literature concerns the locus of associative priming. Associative priming refers to the fact that reaction times are faster to a stimulus if that stimulus is preceded by a stimulus thats tends to co-occur with it in the environment, 

Endogenous and exogenous ERP components

• Exogenous components are those appear to depend on the physical properties of a stimulus (e.g. sensory modality, size, intensity). These have also been called evoked potentials.
• Endogenous components appear to depend on properties of the task (e.g. what the paticipant is required to with the stimulus)

The spatial resolution of ERPs

Inverse problem: the difficulty of locating the sources of electrical activity from measurements taken at the scalp.
The most common way of attempting to solve this inverse problem involves a problem called dipole modelling. This required assumptions to be made about how many regions of the brain are critical for generating the observed pattern of scalp potentials.

MAGNETOENCEPHALOGRAPHY

All electric currents, including those generated by the brain, have an associated magnetic field that is potentially measurable. As such magnetoencephalography (MEG) can be regarded as a parallel method to EEG that is similar in many regards. The biggest potential advantage of MEG over EEG is that it permits much better spatial resolution in addition to the excellent temporal resolution.

In terms of practicalities, MEG is a more challenging and costly enterprise than EEG.