- •Contents
- •1 Introduction
- •A user’s guide?
- •Brain organisation
- •Why is the cerebral cortex a sheet?
- •Cortical origami
- •Does connectivity predict intelligence?
- •Analysis techniques: mapping the brain
- •Structural imaging
- •Functional imaging techniques: PET and fMRI
- •What is the relationship between blood flow and neural activity?
- •The resolution problem
- •Measuring brain activity in real time: MEG and EEG
- •Transcranial magnetic stimulation (TMS)
- •Summary of key points
- •2 The eye and forming the image
- •What is the eye for?
- •Light
- •The structure of the eye
- •Focusing the image
- •The development of myopia
- •Clouding of the lens (cataracts)
- •Photoreceptors
- •Transduction
- •The calcium feedback mechanism
- •Signal efficiency
- •The centre-surround organisation of the retina
- •Light adaptation
- •Duplicity theory of vision
- •Sensitivity, acuity and neural wiring
- •Summary of key points
- •3 Retinal colour vision
- •Why do we need more than one cone pigment?
- •Trichromacy
- •The genetics of visual pigments
- •The blue cone pigment
- •Rhodopsin and retinitis pigmentosa
- •Better colour vision in women?
- •Three pigments in normal human colour vision?
- •The evolution of primate colour vision
- •What is trichromacy for?
- •Summary of key points
- •4 The organisation of the visual system
- •Making a complex process seem simple
- •The retina
- •The lateral geniculate nucleus (LGN)
- •The primary visual cortex (V1)
- •Visual area 2 (V2)
- •Visual area 4 (V4)
- •Visual areas 3 (V3) and 5 (V5)
- •The koniocellular pathway
- •The functional organisation
- •Perception vs. action
- •Blindsight
- •Summary of key points
- •5 Primary visual cortex
- •The visual equivalent of a sorting office?
- •Segregation of layer 4 inputs
- •Cortical receptive fields
- •Spatial frequency
- •Texture
- •Direction selectivity
- •Colour
- •Modular organisation
- •Summary of key points
- •Variations on a theme
- •Monocular or binocular deprivation
- •Image misalignment and binocularity
- •Image misalignment in humans
- •Selective rearing: manipulating the environment
- •Impoverished visual input in humans
- •The critical period
- •What we see, shapes how we see it
- •Summary of key points
- •7 Colour constancy
- •The colour constancy problem
- •The Land Mondrian experiments
- •Reflectance and lightness: the search for constancy in a changing world
- •The biological basis of colour constancy
- •Colour constancy and the human brain
- •Summary of key points
- •8 Object perception and recognition
- •From retinal image to cortical representation
- •Early visual processing
- •A visual alphabet?
- •Complex objects in 3-D: face cells
- •Functional divisions of face cells: identity, expression and direction of gaze
- •The grandmother cell?
- •Are face cells special?
- •Visual attention and working memory
- •Fine-tuning memory
- •A clinical application?
- •Visual imagery and long-term visual memory
- •Summary of key points
- •9 Face recognition and interpretation
- •What are faces for?
- •Face recognition
- •Laterality and face recognition
- •How specialised is the neural substrate of face recognition?
- •The amygdala and fear
- •The frontal cortex and social interaction
- •Faces as a social semaphore
- •Summary of key points
- •10 Motion perception
- •The illusion of continuity
- •Saccades
- •Suppression of perception during saccades
- •What happens if you don’t have saccades?
- •How to stabilise the visual world
- •Navigating through the world: go with the flow?
- •Going against the flow?
- •The neural basis of motion detection
- •Human V5
- •Summary of key points
- •11 Brain and space
- •The final frontier
- •Oculomotor cues
- •Interposition
- •Relative size
- •Perspective
- •Motion parallax
- •Stereopsis
- •The neural basis of three-dimensional space representation
- •The problem of visual neglect
- •The neural basis of neglect
- •Summary of key points
- •12 What is perception?
- •Putting it all together
- •Neuronal oscillations
- •How else to solve the problem
- •What is perception?
- •Change blindness
- •Perceptual rivalry
- •The illusion of perception
- •Summary of key points
- •References
- •Index
14 I N T R O D U C T I O N
about what is going on within this area. How can we overcome this problem and improve the spatial resolution of this technique?
Consider what happens when there is cortical activity in a brain area triggered by a visual stimulus. There are two main changes in the relative deoxyhaemoglobin levels. Firstly, there is an increase in oxygen consumption, caused by an increase in the oxidative metabolism of the stimulated neurons, which leads to an increase in the levels of deoxyhaemoglobin. This happens within 100 ms of the neural activity starting (Malonek & Grinvald, 1996; Vanzetta & Grinvald, 1999), and seems to be localised around the neural activity in a cell column (Grinvald, Slovin & Vanzetta, 2000). The higher deoxyhaemoglobin levels and the increased production of metabolites lead to an increased blood flow to the active region. This produces the second change in relative deoxyhaemoglobin levels, which occurs about 0.5 to 1.5 seconds after the onset of electrical activity for reasons discussed above. This second change in deoxyhaemoglobin levels is far larger than the first, because the increased blood flow overcompensates for the reduction in oxyhaemoglobin levels. Most fMRI systems use a magnetic field of around 0.5 to 1.5 teslas, and detect the second larger change in deoxyhaemoglobin levels, which, as we have seen, is less spatially localised in the cortex and is not closely linked in time with the underlying neural activity. However, the introduction of new high-field fMRI systems (4.7 to 9.4 teslas) means that it is starting to become possible to detect the first change in deoxyhaemoglobin levels, which allows much better temporal and spatial resolution and it is possible to directly measure the activity in single cell columns (e.g. Kim, Duoung & Kim, 2000).
Measuring brain activity in real time: MEG and EEG
Measuring blood flow allows us to track changes in brain activity only indirectly. The blood flow changes occur over a period of seconds, whereas the function of neurons is measured in milliseconds. There are two non-invasive ways to try and measure brain activity in real time: Electroencephalography (EEG) and Magnetoencephalography (MEG). EEG uses electrodes placed on the scalp to measure changes in gross electrical activity across the underlying brain. The electrical signal is sampled simultaneously at multiple locations, usually using 32 or 64 electrodes. One problem with this technique is that, as the electrical activity is being measured through the skull and scalp by electrodes on the surface of the head, the signal may be distorted by its passage through the intervening tissue. MEG does not suffer this problem, as it measures the tiny changes in the magnetic field that accompany electrical currents in the brain.
The passage of any electrical current (such as neuronal activity) induces a magnetic field, which changes as the underlying electrical activity changes. MEG measures the magnetic field induced by neural electrical activity, and so allows underlying neural function to be
T R A N S C R A N I A L M A G N E T I C S T I M U L A T I ON ( T M S ) 15
deduced. The MEG signals are detected by superconducting quantum interference device (SQUID) sensors. As in EEG, the signal is sampled simultaneously at multiple locations. This is accomplished by fitting over 300 SQUID sensors into a helmet-like area that covers the head (Hari, Levanen & Raji, 2000).
The MEG signals that are produced by neural activity are very small, usually in the femto tesla (10 15 tesla) range. To put this figure into perspective, the magnetic field of the Earth is in the order of 0.5 10 4 tesla. So, to avoid contamination of the MEG signal by noise originating from outside sources (such as electrical instruments and power lines), the recordings are carried out typically in a magnetically shielded room.
Although EEG and MEG are able to track changes in neural activity in real time (i.e. with high temporal resolution), they have comparatively poor spatial resolution. As a result, to get the best results, it is becoming common to pair MEG with fMRI, thus allowing good temporal and spatial resolution.
Transcranial magnetic stimulation (TMS)
Functional imaging can only establish the association between task performance and a pattern of cortical activity. However, by using transcranial magnetic stimulation (TMS) to inactivate a particular cortical area transiently, it is possible to test the causal link between activity in a region and a particular behaviour.
TMS is based on the principle of electromagnetic induction. A burst of electric current flowing through a coil of wire generates a magnetic field. If the amplitude of this magnetic field changes over time, it will induce a secondary current in any nearby conductor (Pascual-Leone et al., 1999). The size of the induced current will be dependent on how fast the magnetic field changes its size. In TMS, a magnetic coil is held over the subject’s head and, as a brief pulse of current is passed through it, a magnetic field is generated that passes through the subjects scalp and skull. This time-varying magnetic field induces a current in the subject’s brain, and this stimulates the neuronal tissue. The neurons within the stimulated area fire off in a burst of activity, followed by a period of inactivity while the neurons recover. During the recovery period, it is assumed that the sensory or cognitive functions normally performed by this area will be temporarily deactivated, allowing the experimenter to deduce the function normally undertaken by the area. In many experiments, single pulses of stimulation are used. In others, a series of pulses at rates of up to 50 Hz (this called repetitive TMS or rTMS). This latter procedure can be dangerous and can cause seizures, and so must be used with caution. With this caveat, TMS is an extremely useful technique. It allows investigators to test reversibly whether a particular area undertakes the functions ascribed to it, and has been widely used in exploring visual perception in human observers.
16 I N T R O D U C T I O N
Summary of key points
(1)The human cortex is a strip of neurons, usually divided into six layers, which vary in thickness from about 1.5 to 4.5 mm. To fit it all into the confines of the skull, the cortex is thrown into a series of ridges (gyri), and furrows (sulci).
(2)Why is the cortex a flat sheet? The radial-unit hypothesis suggests that increasing cortical size is based on increases in the number of radial columnar units that make up the cortex, rather than changing the number of cells in each column.
(3)The cortical sheet folds in specific places. The folding is designed to minimise total axon length. In an outwards fold the ridge is directed away from the white matter and the brain interior, and the length of axonal connections between the two banks of the fold is small. Such folds could bring together heavily connected areas. In an inwards fold, the crease is directed towards the white matter and so the white matter distance between the two folds is long. Inwards folds thus should tend to fall between sparsely connected areas.
(4)The cortex is divided into four main lobes: the occipital lobe, the temporal lobe, the parietal lobe and the frontal lobe. These lobes are then subdivided into different functional areas.
(5)There are several rules of organisation for brain organisation. Firstly, neurons with similar patterns of connections and response properties are clustered together to form areas. Secondly, these different areas themselves are subdivided into smaller processing units. Thirdly, corresponding higher visual areas on different sides of the brain do slightly different jobs, a process called lateralisation.
(6)Computerised tomography (CT), or computer assisted tomography (CAT), uses X-rays for a non-invasive analysis of the brain. The X-ray emitter and detector scan the head from front to back, and the process is repeated until the brain has been scanned from all angles. The computer takes the information and plots a twodimensional picture of a horizontal section of the brain. The patient’s head is then moved up or down, and the scan is taken of another section of the brain.
(7)Magnetic resonance imaging (MRI) passes an extremely strong magnetic field through the patient’s head; this causes the nuclei of some molecules to emit radiowaves. Different molecules emit energy at different frequencies. The MRI scanner is tuned to detect the radiation from hydrogen molecules. Because these molecules are present in different concentrations in different brain tissues, the scanner can use the information to prepare pictures of slices of the brain.
(8)Positron emission tomography (PET) measures the flow of radioactively labelled blood to different areas of the brain. It is assumed that an increased flow to a brain area is an indication of increased
S U M M A R Y O F K E Y P O I N T S 17
function. A more accurate measure of blood flow is functional magnetic resonance imaging (fMRI), which is a refinement of the MRI technique. This technique takes advantage of the differences in the magnetic profile of oxygenated and deoxygenated haemoglobin to map bloodflow and the metabolic use of oxygen in the brain.
(9)Electroencephalography (EEG) measures the electrical activity of the brain through electrodes placed on the scalp. Magnetoencephalography (MEG) measures the electrical activity of the brain indirectly by measuring the changes in magnetic field induced by the fluctuations in the brain’s electrical activity. These are both functional imaging techniques and, although they have good temporal resolution, they have relatively poor spatial resolution.
(10)In Transcranial magnetic stimulation (TMS) a magnetic coil is held over a volunteer’s head and a brief pulse of current is passed through it; a magnetic field is generated that passes through the subject’s scalp and skull. This time-varying magnetic field induces a current in the subject’s brain, and this stimulates the neuronal tissue and leads to its temporary deactivation.
