Ongoing spontaneous activity can be recorded at the surface of the human head using MEG or EEG. Such activity typically is characterized by regions of relatively large amplitude oscillatory patterns that vary as a function of position on the head and state of the subject. Pathological responses such as certain forms of slow oscillation or the spikes and waves associated with epileptic activity often can be clearly resolved in the ongoing EEG record. The signals associated with responses to individual stimuli or other punctate cognitive or control processes typically are much smaller and require specialized experimental paradigms and signal processing techniques to pull the signals out of the noise.
Brief presentations of sensory stimuli elicit a sequence of responses in the chain of specialized cortical and subcortical areas that process sensory information. In the primate visual system, over three dozen areas have been identified that are involved in the processing of visual information. Other sensory modalities such as auditory and somatosensory systems involve smaller numbers of areas and less cortical real estate but still employ distributed processing strategies.
In order to enhance the consistent aspects of the sensory response while suppressing the contribution of other physiological processes or environmental noise, most investigators employ averaging of temporal sequences time-locked to the stimulus. In this sort of evoked response paradigm, individual stimuli are typically presented in isolation. Different examples of a class of stimuli are often presented in a random sequence. The interstimulus interval typically is varied within some limits to minimize habituation and thwart the generation of temporal expectations by the subject;
such effects can influence the amplitude and timing of certain components of the response. The time course of the electrophysiological response to a single stimulus ranges from tens of milliseconds to a few hundred. Thus, by presenting stimuli at intervals of 500 msec or more, it is possible to examine the entire time course of the response with little or no overlap from preceding or subsequent responses. Figure 4 illustrates an example of some of the techniques used for visualizing a somatosensory evoked response.
An alternative approach for evoked response paradigms is to present stimuli at high rates, accepting the response overlap. Instead of discrete stimulus presentations, such paradigms often impose rapid changes on an ongoing stimulus, such as amplitude modulation of a continuous tone or contrast reversal of a patterned video display. Because the response does not return to baseline between stimuli (i.e., residual activation is maintained between trials), this class of techniques is referred to as steady-state methods. However, when using typical AC-coupled recording methods, the response must be modulated in order to be detectable. The time-locked response is typically isolated using Fourier transform techniques. The amplitude at the stimulation frequency is taken as a measure of the evoked response, whereas the phase is taken as a measure of the temporal delay (or latency) of the response.
The primary advantage of this strategy is speed; steady-state methods provide an efficient way of collecting and analyzing topographic data produced by MEG or EEG. However, such methods also have disadvantages. Fast stimulus presentations produce a measure of sensory overload, often leading to habitua-tion or reduced levels of attention to stimuli. Steady-state methods may also introduce phase ambiguities. For example, visual evoked responses show evidence of an initial activation in layer 4 of the primary cortical visual area, V1, followed by activation in other layers and eventually by feedback activation in V1 from higher visual areas. At high stimulation rates the temporal relationship between various phases of the V1 cortical response may be obscured. Further, the subsequent activation of other visual processing areas may produce field or potential topographies that overlap with the responses of earlier areas. The loss of timing information removes an important tool for the identification of sources and for studies of the dynamics of information processing.
Experimental studies employing MEG have demonstrated a clever application of high-frequency stimulation techniques. By modulating the stimulus at frequencies that may be too high to consciously perceive, it is possible to frequency-tag the downstream response. Such methods can be used to tag the hemifield of stimulation in a wide-field visual stimulus or to identify the stimulated ear in a dichotic listening paradigm. Residual modulation at the tag frequency can be used to identify the origin of a response even after the arrival of the signal at a higher sensory processing area with convergent bilateral inputs.
By presenting fast pseudo-random sequences of individual stimuli, it is possible to achieve much of the efficiency of steady-state techniques while avoiding several of the problems. For example, a video display can be divided into elements that are turned on and off in an apparently random sequence, such as an m-sequence. The temporal activation sequences are designed to be orthogonal, i.e., each element has its own unique activation sequence. This allows correlation techniques to be used to extract the spatial and temporal patterns of response to each element of the display. This method can be very efficient because many stimulus elements can be presented simultaneously. However, this leads to stimuli that are decidedly nonphysiological and may be a bit disconcerting. At present, the method appears to be more useful for rapid mapping of the systematic parametric organization of primary sensory areas (i.e., retinoto-pic, tonotopic, or somatotopic projections) than as a tool for probing higher cognitive processes.
Sensory evoked response paradigms provide a powerful and robust tool for probing the functional architecture and dynamics of the neural systems devoted to the processing of sensory information. However, this is only one of several classes of functional activity within the human brain.
The control of voluntary movement is another critical capability of higher organisms. Although the coordination and fine-tuning of movement appear to be distributed through several brain centers, the initiation of movement is a function of the motor cortex, located adjacent to primary somatosensory areas in humans. Some investigators have used cued trials to study motor function, hoping that the temporal jitter in the reaction time does not wash out the targeted motor response. This strategy also produces a time-locked response to the sensory cue, which can interfere with the analysis of the motor response. An alternative strategy is to use self-paced tasks and to time-lock averaging to the motor response. For simplicity, the response can be a button press or similar action that is readily converted into an electronic timing signal. A bit more sophisticated strategy is to base experimental timing on a measured physiological response. For example, it is possible to record an electromyogram signaling the activation of specific muscle groups by using the same basic technology as EEG. These are large, robust responses that can be detected in a continuous recording, often using simple threshold techniques. This sort of activity, which may not be evoked directly by external stimulation but is associated with externally observable consequences, is referred to as an event-related response. In addition to motor processes or simple behavioral trials such as reaction time detection tasks, event-related experimental designs are often used to probe high order cognitive processing activity. In some cases, the "event" is only apparent as an internal state, e.g., conjunction of a particular stimulus with a specific behavioral or cognitive task.
The general strategy for event-related cognitive studies is to employ a set of tasks designed to isolate and contrast the processes of interest. Measures derived from MEG or EEG are used as an index of activation and, thus, of the underlying cognitive processes. Such studies often employ well-balanced control trials, which account for sensory or motor components of a response while manipulating the relevance or difficulty of the cognitive component. Such strategies have been used to isolate and probe various aspects of selective attention. For example, many designs used for attention studies employ a cue stimulus presented before the probe trial to direct attention to one region of the visual field or another. The subject might be instructed to respond to a particular type of stimulus only when presented at the attended site. Thus, in the same experiment, a given physical stimulus might serve as the response target, an inappropriate stimulus at the attended site, or an irrelevant stimulus that can be ignored.
Other designs tap the endogenous cognitive skills of the subject. For example, a list of real words might be presented visually along with interspersed pseudo-words (i.e., pronouncible constructs that look like words but have no meaning) or nonwords. The nature of the observed response varies as a function of the
Was this article helpful?