Info

Figure 4.2. A schematic representation of ERP components elicited by auditory, infrequent target stimuli. The three panels represent three different voltage x time functions: the left bottom panel shows the very early sensory components (with a latency of less than 10 ms); the left top panel shows the middle latency sensory components (with a latency of between 10 and 50 ms); and the right panel shows late components (latency exceeding 50 ms). Note the different voltage and time scales used in the three panels, as well as the different nomenclatures used to label the peaks (components). (Adapted with permission of the author from Donchin, 1979, with kind permission of Springer Science and Business media.)

Although the exogenous/endogenous distinction provides a useful method for classifying many ERP components, there are potentials that possess characteristics that are intermediate between these two groups, and are therefore called "mesogenous." The N100 (see Section 4.3.1) is such an example, as it is sensitive to both the physical properties of the stimulus and the nature of the interaction between the subject and the event (e.g., whether the event is to be attended).

2.3. From the brain to the scalp: the generation and physiological basis of ERPs

In this section, we review evidence that relates the scalp-recorded electrical activity to its underlying anatomical and physiological basis (see also Allison, Wood, & McCarthy, 1986; Nunez, 1981). It is generally assumed that ERPs are distant manifestations of the activity of populations of neurons within the brain. This activity can be recorded on the surface of the scalp because the tissue that lies between the source and the scalp acts as a volume-conductor. Because the electrical activity associated with any particular neuron is small, at the scalp it is only possible to record the integrated activity of a large number of neurons. Two requirements must be met for this integration to occur: (a) the neurons must be active synchronously, and (b) the electric fields generated by each particular neuron must be oriented in such a way that their effects at the scalp cumulate. As a consequence, only a subset of the entire brain electrical activity can be recorded from scalp electrodes.

Two considerations further restrict the likely sources of the scalp-recorded ERP. First, because the ERP represents the synchronous activity of a large number of neurons, it is probably not due to the summation of pre-synaptic potentials (action potentials or spikes), because these potentials have a very high frequency and short duration. In contrast, post-synaptic potentials, having a relatively slower time course, are more likely to be synchronous, and therefore to summate to produce scalp potentials. Thus, it is commonly believed that most scalp-recorded ERPs are the outcomes of summation of post-synaptic potentials of a large number of neurons that are activated (or inhibited) synchronously (see Allison et al., 1986).

A second consideration concerns the orientation of neuronal fields. Because the electric fields associated with the activity of each individual neuron involved must be oriented in such a way as to cumulate at the scalp, only neural structures with a specific spatial organization may generate scalp ERPs. Lorente de No (1947) specified the spatial organizations that are required for the distant recording of the electrical activity of a neural structure. He distinguished between two types of configurations: "open fields" and "closed fields." A structure having an open-field organization is characterized by neurons that are ordered so that their dendritic trees are all oriented on one side of the structure, whereas their axons all depart from the other side. In this case, the electric fields generated by the activity of each neuron will all be oriented in the same direction and summate. Only structures with some degree of open-field organization generate potentials that can be recorded at the scalp. Open fields are obtained whenever neurons are organized in layers, as in most of the cortex, parts of the thalamus, the cerebellum, and other structures.

A structure with a closed-field organization is characterized by neurons that are concentrically or randomly organized. In both cases, the electric fields generated by each neuron will be oriented in very different, sometimes opposite, directions, and therefore will cancel each other out. Examples of closed-field organization are given by some midbrain nuclei.

From this analysis it is clear that ERPs represent just a sample of the brain electrical activity associated with a given event. Thus, it is entirely possible that a sizeable portion of the information processing transactions that occur after (or before) the anchor event is "silent" as far as ERPs are concerned. For this reason, some caution should be used in the interpretation of ERP data. For instance, if an experimental manipulation has no effect on the ERP, we cannot conclude that it does not influence brain processes. By the same token, if two experimental manipulations have the same effect on the ERP, it cannot be concluded that they necessarily influence completely identical processes.

2.4. From the scalp to the brain: inferring the sources of ERPs

So far we have examined how particular properties of neuronal phenomena may determine whether they will be recorded at the scalp. We have approached the problem of ERP generation in the "forward" direction - from properties of the generators to predictable scalp observations. In most cases, however, we have only limited information about the neural structure(s) responsible for a specific aspect of the ERP. The typical ERP database consists of observations of voltage differences between scalp electrodes or between scalp electrodes and a reference electrode. To determine which neural structures are responsible for the scalp potentials (i.e., to identify the neural generators of ERPs) we must solve the "inverse problem" - that is, we have to infer the unique combination of neural generators whose activity results in the potential observed at the scalp.

In solving this problem, we are confronted with an indefinite number of unknown parameters. In fact, an indefinite number of neural generators may be active simultaneously, and each of them may vary in amplitude, orientation of the electric field, and location inside the head. Because a limited number of observations (the voltage values recorded at different scalp electrodes) is used to estimate an indefinite number of parameters, it is clear that the inverse problem does not have a unique solution (i.e., an infinite number of different combinations of neuronal generators may produce the same scalp distribution). A further complication is that the head is not a homogeneous medium. Therefore the propagation of an electric field generated by the activity of a given structure is difficult to compute. A particularly important distortion of the electric fields is caused by the skull - a very low conductance medium that reduces and smears electric fields. For all these reasons, we cannot unequivocally determine which structures are responsible for the ERP observed at any point in time, when the only information available is given by the potentials recorded at scalp electrodes.

Notwithstanding these problems, investigators have tried to identify the neural sources of scalp-recorded ERPs using a variety of approaches, involving both noninvasive and invasive techniques. Noninvasive techniques include:

(a) scalp recordings from dense electrode arrays combined with interpolated mapping and source analysis algorithms (which involve complex mathematical procedures and are based on a number of assumptions); and (b) combining ERP recordings with other imaging methods that possess higher spatial resolution (e.g., PET, fMRI, MEG, EROS), to help restrict the number of solutions to the inverse problem. Invasive techniques include: (a) recordings from indwelling macro-electrodes (in humans or animals); and

(b) lesion data (also in humans or animals).

Was this article helpful?

0 0

Post a comment