normally in everyday situations, but in an experimental setting it was possible to study situations where each brain half was unaware of the sensory input presented to the other half. By presenting different, or even conflicting, information to each half, it was possible to study differences in left and right brain information processing.

If, for instance, the left hand was holding a cup behind a curtain so that it could not be seen, and the patient was asked what he was holding in his hand, he would be unable to answer. The explanation given was that the speech center, being located in the left brain, had insufficient information about what the left hand was doing. However, the patient was able, with his left hand, to write the correct answer ‘cup’ (since the visual information was passed through to the right half of the brain). When the word ‘book’ was presented only to his left field of view, the patient would be unable to convey orally what he had read although he was able to write the answer with his left hand. When the patient was forced to respond orally, the answer would be based on a guess. However, after he had answered (with his left brain), the right brain would hear the answer and decide whether it was right or wrong. In other words, communication within each brain half was intact, but the two halves could not communicate with each other internally. The two brain halves operated as individual, independent units.

There is an anecdote about another ‘split brain’ patient who could not manage to get dressed for a party. The reason, it was said, was to be found in a conflict between the two brain halves. One of them wished to go to the party, but the other did not like the idea, so while one hand was dressing the other hand was undressing.

Localization of brain activity: methods

Until now, the method that has given us the most information by far about the behavior of nerve cells is the use of microelectrode recordings from single units. A few other recent methods provide information about the functioning of cell groups, the selective activation of which is achieved through a specific design of the physical stimulation.

Optical imaging

In this method, the cortical surface is observed while it is illuminated with red light. Active cortical regions absorb more light than the less active ones. The temporal and spatial changes in light absorption can be recorded with a video camera as a means of mapping cortical activity. The distribution of orientation specific neurons of Figure 8.4(b) was obtained using this technique. The method can be applied to experimental animals and to human patients undergoing neurosurgery.

Positron emission tomography

Positron emission tomography (PET) is a method used to localize brain activity that is linked to a particular task. It takes advantage of the fact that cells that are active in performing a particular task have a greater energy uptake than when they are at rest. After injecting a small dose of radioactive sugar (glucose) into the blood, it is possible to register which brain areas are more radioactive (emitting gamma rays) and therefore have a higher concentration of glucose than the surroundings. Cortical areas that have been most active during the task will have consumed the most glucose and will ‘lighten up’ in the PET scan maps, regardless of whether the task was a motor activity, such as moving the index finger, a sensory task or a mental one. In this way it has been shown that emotions and cognitive (logical, lingual) functions are localized to certain areas in the brain. The spatial resolution in PET scans is a few millimeters. The time resolution is fairly low and rapid changes cannot be registered.

Magnetic resonance imaging and functional magnetic resonance imaging

Magnetic resonance imaging (MRI) is a means of imaging the anatomy of the brain that is often used in conjunction with a PET scan. MRI provides a clear outline of the brain, with markers that are needed for determining the precise position of PET activity. MRI is based on the fact that electrons are influenced by an external magnetic field; they are like small magnets that are forced to ‘line up’ in a certain direction in the magnetic field. This alignment can be registered by radio waves.

MRI today has a spatial resolution of less than 1 mm, which is considerably better than PET. The temporal resolution is less than 1 s and is continuously improving. Further developments of the MRI technique have made it possible to localize brain activity by functional MRI (fMRI, Raichle, 1994). Brain activity is visually mapped by subtracting the MRI image of an idle brain from the image of the brain actively engaged in a visual or mental task, for example. The method relies on the fact that the magnetic properties of blood depend on its oxygen content, and in fMRI it is the oxygen uptake in the blood that is monitored. In fMRI the blood oxygen level is the measure of neural activity, and changes in activity in a defined brain region can be registered by fMRI.

Sometimes the level of activity in an fMRI signal can be used to produce contours of equal activity when the stimulus is changed. By recording a contrast–response function, it is possible to compare a threshold activity with psychophysical sensitivity for the same stimulus and the same subjects. However, because the resting brain has some baseline activity, only the activity change induced by changing stimulus contrast or by introducing a new attribute is of interest. A typical difference in the fMRI signal is only about 5 percent of the total signal, and fMRI sensitivity thus lags far behind psychophysical sensitivity.

Figure 8.10 The two visual areas V5 are activated by movement of a visual stimulus. Here, areas are localized by means of fMRI during central fixation. Movement-dependent activity is indicated by the symmetrically located white areas half away from the midline. The smaller white spots at the back of the brain, closer to the midline, are from V1/V2 (Freitag et al., 1998).

Figure 8.10 shows fMRI images of two subjects looking at moving targets. The white areas are those that were particularly active during this task. The two large areas on each side of the brain are areas V5, where the cells are highly specialized for movement detection.

A shortcoming of PET and fMRI is their poor temporal resolution. Better temporal resolution can be obtained by measuring electromagnetic activity in the brain. However, this can be done non-invasively only at the cost of spatial resolution.

Visual evoked potentials

A Visual evoked potential (VEP) recording is a form of electroencephalogram (EEG) for measuring the potential differences that occur over the brain during visual processing. In the simplest arrangement, this is done by placing one or more electrodes on the skin above the visual cortex and another at some reference position elsewhere on the head. With this arrangement it is possible to measure the fluctuations in electric potential that correlates with activity in area V1 and adjacent areas. With a large number of electrodes (more than 100) distributed over the scalp in a systematic fashion, it is possible to record the topographical distribution of the fluctuating electric potential arising from visual stimulation, or from other sensory input. Using a sophisticated model of the electric properties of the head, and an MRI image of its anatomy, the data can be analyzed mathematically to determine the location of cell groups generating that particular topography of electric potential across the skull (brain mapping). Area V1 is relatively accessible for VEP recordings, although part of the fovea is hidden between the two brain halves (see Figure 8.3). Cortical processing of visual information is distributed over large parts of the brain, and the different attributes of a stimulus may activate several brain areas simultaneously.

One advantage with the multielectrode VEP method over PET is that it does not interfere with the ‘inner life’ of the brain and that it does not require radioactive marking. The spatial resolution in VEP brain-mapping depends on the number of electrodes and can be better than with PET. However, it is less reliable for sources of activity that are situated in the deeper layers of the cortex. VEP has the best time resolution of all techniques. Resolution is determined by sampling frequency and can be set to values below 1 ms. Various sources of noise (external noise from the mains, electrical instruments and machines, and internal noise from global EEG and muscle activity) must be minimized when recording VEP. Because cortical activity unrelated to visual stimulation (EEG) is significant, even in differential recordings using a scalp reference, a VEP recording is usually built up of many repeated responses to identical presentations of the same stimulus. Since the timing of the response contribution from non-visual brain activity is random with respect to the stimulus, the electric noise from different sources will largely cancel out when a large number of synchronized responses are added together. The visual responses, however, are time-locked to the stimulus, and will survive this averaging procedure.

Figure 8.11 shows a conventional experimental setup for recording VEPs by using three electrodes. Potential differences between the forehead and the V1 region at the back of the head are registered and stored on a computer; the ear is connected to ground. A combination of the methods mentioned above, each of which is continuously being developed and improved upon, is a powerful tool in brain research.

Figure 8.11 An experimental setup for recording VEP. The potentials are, for example, measured between the electrodes Er and Em.