Neurophysiological and Psychophysiological Approaches to Schizophrenia and Its Pathogenesis

Keith H. Nuechterlein, Ph.D. and Michael E. Dawson, Ph.D.

The striking abnormalities of perception, thinking, and speech that occur during active psychotic periods in schizophrenia are likely due to alterations in the normal neural control of information processing (12,84). Negative symptoms of schizophrenia, such as flat affect, apathy, and poverty of speech (3), seem to entail decreased integration of emotional and motivational functions with perceptual and cognitive functioning. Because clinical symptoms are several steps removed from direct neuropsychopharmacological effects, neurophysiological and psychophysiological methods can provide critical intermediate steps in our understanding of the progression from neurochemical abnormalities to the clinical symptoms of schizophrenia. These intermediate steps may also provide endophenotypes that simplify the search for genetically transmitted components of vulnerability to schizophrenia (43,60,84). We will describe several prominent methods, a sampling of results, and sources for further information.

One prominent psychophysiological dysfunction in schizophrenia that is not reviewed here is eye tracking dysfunction, which is among the most promising indicators of genetic vulnerability to schizophrenia. Fortunately, excellent recent reviews of the extensive literature on eye tracking abnormalities (e.g.,59,63,74). Aspects of neurocognitive performance that may also reflect critical underlying factors in vulnerability to schizophrenia are reviewed in Treatment-Resistant Depression ) and elsewhere (5,12,82,84).

ELECTROENCEPHALOGRAM AND EVENT-RELATED POTENTIAL MEASURES

Both the electroencephalogram (EEG) and event-related potentials (ERPs) have been employed extensively in studies of schizophrenia. The EEG reflects a combination of excitatory and inhibitory postsynaptic electrical potentials recorded between pairs of electrodes on the scalp. The neocortex that lies near the electrodes gives rise to these potentials, but the potentials also reflect influences from ascending pathways from limbic, thalamic, and reticular nuclei. The EEG is a physiological measure of the momentary functional state of the underlying cerebral structures, although it may be affected by any neuroanatomical abnormalities. Shagass (99) has noted that the critical advantages of the EEG for psychiatric research are that it is 1) noninvasive and radiation-free, 2) repeatable, 3) closer to brain effects for pharmacological studies than blood levels, 4) multidimensional (frequency, amplitude, wave symmetry, and the temporal and spatial aspects of each), and 5) extremely sensitive to changes in alertness and cognitive activity. On the other hand, its limitations are that 1) the usual scalp recordings index primarily surface cortical activity, 2) the potentials include attenuation by tissue conduction, 3) recordings involve some overlap between potentials that originate at different locations, 4) care must be taken to eliminate artifacts from electrical potentials from noncerebral sources, and 5) the subject's state of alertness and cognitive activity need to be considered and possibly controlled when a psychopathological characteristic is the object of study (99).

The typical EEG scoring procedure involves power spectral analyses, which examines the amount of electrical activity in delta (<4 Hz), theta (5–7 Hz), alpha (8–13 Hz), and beta (14–30 Hz) frequency bands. Processing of EEG recordings has advanced substantially in recent years. One procedure involves examination of EEG coherence, which refers to the similarity of EEG waveform components at spatially separate regions of the scalp. Coherence is assumed to reflect functional connections among underlying brain regions during a given form of mental activity (115). Topographic mapping of the power or amplitude values across the surface of the head is also increasingly used to examine the regional distribution of electrophysiological activity. Another analytic technique based on nonlinear dynamics examines the dimensional complexity of the EEG, treating the EEG as "chaotic" time series behavior. Dimensional complexity refers to the number of parameters that contribute to time series behavior (57,91).

Other current methods of examining the physiological responses of the brain, such as positron emission tomography (PET), regional cerebral blood flow, and functional magnetic resonance imaging, offer better spatial resolution than ERP methods. However, the superior temporal resolution of ERP methods, combined with increasing spatial resolution, continues to offer distinct advantages for studies that seek to understand how regional brain activity develops and changes over very short periods of time in sensory, perceptual, and cognitive tasks.

The ERP temporal components can be used to detect successive aspects of registration and processing of a stimulus. Auditory ERP components are often divided into brainstem components (first 10 msec), middle latency components (10 to about 90 msec), and long-latency components (100 msec and later) [36,38]. Starting with the positive-polarity response at about 50 msec, the components are labeled by positive or negative polarity (P or N) and by their latency in msec (e.g., P50, N100). The ERP components through 200 msec have usually been called exogenous or stimulus-driven, because they mainly reflect particular physical stimulus properties of the stimulus. The components after 200 msec are referred to as endogenous, concept-driven, or context-dependent, because they are strongly influenced by internal stimulus representations and the role or significance of the stimulus as defined by the task and the subject (36,38,64).

The EEGs of schizophrenic patients have been found to differ from those of normal subjects in a variety of ways, some of which may reflect the level and type of psychiatric symptoms that the patient is experiencing at the time of assessment. One prominent finding is an increased frequency of EEG records reflecting high activation levels, as judged by reactivity of the EEG upon opening the eyes, among schizophrenic patients in an unmedicated state (99). Consistent with hyperactivation are quantitative results showing that EEGs of schizophrenic subjects have lower alpha power, increased variability of frequency, and higher wave symmetry than comparison subjects (99). Given that excesses in both slow and fast EEG activity have been reported, however, Shagass (99) suggests that dysregulation of brain activity in schizophrenia might be a more basic problem than hyperactivation per se.

Koukou and her colleagues have addressed more directly than most investigators the issue of whether EEG abnormalities represent trait-like vulnerability to schizophrenia or a characteristic of a schizophrenic psychotic state. Koukou and Manske (70) selected schizophrenic subjects who were either in an unmedicated state with a recent onset of positive symptoms or who had shown a full symptomatic remission after a first episode and had been medication-free for three months. Abnormal power in low frequency bands (2-8 Hz) in reaction to auditory stimuli was found to be present only in the symptomatic schizophrenic patients and therefore is apparently a reflection of clinical state. Possible indicators of a vulnerability trait were also suggested but need evaluation in longitudinal studies across clinical states before any clear conclusions can be drawn.

Several topographic EEG studies have found evidence of excessive slow activity (delta) in frontal areas (e.g., 48,80). This is conceptually consistent with PET and regional cerebral blood flow evidence of reduced frontal activity in at least some schizophrenic patients and increased frontal delta activity was correlated within schizophrenic samples with PET hypofrontal metabolic patterns (48). Adding to this picture is the finding that EEG alpha-band coherence between the right prefrontal region and both right inferior temporal and right occipital regions during performance of a sustained attention task with degraded visual stimuli, but not during an eyes-closed resting condition, is lower in schizophrenic patients than in normal subjects (56) . Thus, schizophrenic patients may have decreased functional connectivity in a neural network for identification of visual stimuli that involve prefrontal, inferior temporal, and occipital areas.

Results using dimensional complexity and other measures from chaos theory are beginning to appear and offer promise for detecting individual differences in dynamic brain activity beyond those that have been detected by prior analytic procedures (29). Dimensional complexity of the EEG appears to vary across brain regions in a different pattern in schizophrenic patients than in normal subjects (29,57,69), but, thus far, specific results appear to be dependent on methodological approach. Using Laplacian filtering (to enhance spatial resolution, inclusion of higher EEG frequency ranges that might reflect cortico-cortical communication) and a passive visual stimulation condition, Hoffman, Buchsbaum, and collaborators (57) recently found reduced dimensional complexity in schizophrenia, particularly in the left frontal and right parietal areas. Thus, schizophrenic abnormalities involving frontal lobe activity and functional connectivity to other regions are prominent in recent EEG findings across several specific forms of analysis, consistent with one theme of PET and regional cerebral blood flow findings.

The most common ERP paradigm has examined P300 (P3) amplitude and latency in a situation in which an occasional, unpredictable physical change in a stimulus occurs within a series of identical stimulus presentations (the "oddball" paradigm). The oddball stimulus may, for example, differ from the repeated standard stimulus in pitch or duration. The P300 response in this situation is an index of cortical response related to recognizing and processing the significance of the relatively rare stimulus. Schizophrenic patients were first found to show reduced P300 amplitude in this paradigm a little more than 20 years ago, and reduced P300 amplitude has been among the most robust of ERP results for schizophrenia (32,40,64). Because P300 is believed to be an index of consciously controlled aspects of information processing, this P300 abnormality contributed to the view that controlled information processing is impaired in schizophrenia but more automatic information processing is intact (16). P300 reduction in schizophrenia appears to be linked to the presence of background stimuli that make the identity of the next stimulus uncertain. It occurs even when experimental instructions do not call for motivated effort, and in some studies has been significantly related to the level of negative symptoms (32,89). P300 paradigms have also bolstered the evidence that schizophrenic patients have deficits in maintaining attention and show inappropriate allocation of attentional resources to novel distracting stimuli (32,79). Recent work indicates that a specific aspect of P300 amplitude reduction (smaller amplitude over left compared with right lateral sites) is linked to thought disorder levels and reductions in the volume of the left posterior superior temporal gyrus among schizophrenic patients (76).

Decreased auditory P300 amplitude or increased P300 latency has also been found in several recent studies of first-degree relatives of schizophrenic patients (9, 67, 92!popup(ch117, 95), suggesting that P300 latency or amplitude might be an indicator of a genetic vulnerability factor for schizophrenia. Friedman and Squires-Wheeler (40) argue that these findings might reflect the older age of the first-degree relatives of schizophrenic probands relative to control subjects. Although Blackwood and colleagues (9) found that the relationships between age and P300 latency and amplitude were not significant in the age range from 16–50 years, both were significant in the control group (and also for P300 latency in the schizophrenic group and in the group of nonschizophrenic relatives of schizophrenic probands) when the full included range of 16–60 years was examined. Thus, the possible effect of age on these P300 group differences needs further examination. It is nevertheless of substantial interest that Blackwood et al. (9) found evidence of significant bimodality of P300 latency among nonschizophrenic relatives of schizophrenic probands, with the two means corresponding roughly to those of the normal controls and the schizophrenic patients. This P300 latency abnormality makes a contribution to identification of potential genetically prone relatives independent of that made by smooth-pursuit eye tracking dysfunction. Furthermore, Roxborough et al. (92) found that nonschizophrenic relatives of schizophrenic probands who had abnormally long P300 latencies, compared with relatives with normal P300 latencies, showed impairments in verbal fluency and verbal memory span tasks that are sensitive to frontal and medial temporal lobe dysfunction. Neuropsychological deficits usually associated with frontal and temporal lobe dysfunctions are prominent in schizophrenia, as discussed in Treatment-Resistant Depression).

P300 findings in children at increased genetic risk by virtue of having a schizophrenic parent have been mixed. Friedman et al. (39) found no evidence of P300 abnormalities among the complete initial sample from the New York High Risk Project, and Squires-Wheeler et al. (103) found that P300 amplitude recorded in adolescent offspring of schizophrenic probands did not predict later young adult clinical assessments. On the other hand, Schreiber et al. (96) showed that children of schizophrenic patients had prolonged P300 latencies (although not reduced amplitude) when compared with control children matched for age, sex, and educational level. The excellent review by Friedman and Squires-Wheeler (40) discusses further the methodological issues that need to be addressed in order to determine whether P300 latency or amplitude can serve as a useful vulnerability indicator for schizophrenia.

Another ERP paradigm applied to schizophrenia has focused on endogenous negativity in the waveform associated with effortful or attention-demanding processing of information (104). This paradigm records ERPs that occur to very brief visual letter arrays in a simple reaction time task (without the need to discriminate letters) and compares them to ERPs that occur to the same visual letter arrays when the subject is instructed to discriminate whether one of two letters is included among others in the arrays. The latter situation, called the Span of Apprehension task, has been shown to detect early visual processing abnormalities in schizophrenic patients in remitted and psychotic states. It is hypothesized to be sensitive to a genetic vulnerability factor relevant to schizophrenia (5). By subtracting the ERP for the simple reaction time condition from the ERP for the Span of Apprehension condition, the electrophysiological response associated with the visual discrimination process has been examined (104). Through this subtraction procedure, Strandburg and colleagues have shown that schizophrenic patients, including schizophrenic children, produce a significantly smaller Span elicited negativity (SEN) than normal subjects (104). Although exogenous (stimulus-driven) N1 and P2 components occur in this situation, this early endogenous response differs from the usual exogenous N1 and P2 ERP components in both topography and time course. Figure 2 illustrates the way that topographic field maps can be used to display the differing amplitude and regional distribution of the ERPs and the difference potentials in this paradigm. Strandburg et al. (104) conclude that the smaller Span-elicited negativity in schizophrenia may reflect impaired ability to allocate attentional resources sufficient for the serial search, pattern recognition, and stimulus identification demands of this tachistoscopic visual discrimination task.

An early positive ERP component, P50, has been the focus of substantial recent research as a possible index of a vulnerability factor for schizophrenia. A measure of sensory gating during the initial processing of stimuli can be obtained by examining the extent to which P50 amplitude to a simple auditory stimulus is reduced when another simple auditory stimulus occurs immediately before it (37). The concept of a sensory or sensorimotor gating process has been central to work involving P50 and to work with a paradigm that we discuss in the next section, prepulse inhibition of the startle blink response. The concept of sensory or sensorimotor gating derives from the fact that all of us are constantly bombarded with a multitude of external and internal stimuli as we navigate through the world. For the most part, we are able to select and consciously process those specific stimuli most relevant to our current activities and goals, while somehow screening out, or gating, irrelevant stimuli—often to the point that we are not even consciously aware of them. Some basic aspects of stimulus detection and selection may normally occur relatively automatically, whereas other aspects require active, conscious attention (84).

Acute schizophrenic patients in the early stage of their psychosis often describe vivid experiences suggestive of an impaired ability to gate irrelevant stimuli and focus on relevant stimuli (78). For example, one patient reported, "I am speaking to you now but I can hear noises going on next door and in the corridor. I find it difficult to shut these out and it makes it more difficult for me to concentrate on what I am saying to you" (78, p. 112). McGhie and Chapman (78) hypothesized that a breakdown in the selective-inhibitory function of attention occurs in schizophrenia, leading to a sense of being flooded with incoming sensory data. The notion of an impaired sensory filter, or sensory gate, in schizophrenia remains a popular one today (12,13 37). One key contemporary issue is whether such an impairment is due to faulty automatic processing or faulty controlled (or attention-demanding) processing of the incoming stream of stimuli.

In the P50 conditioning-testing paradigm, two brief auditory stimuli (usually clicks) are presented in rapid succession, typically 500 msec apart. Series of these paired stimuli are presented, usually at 10-sec intervals. The P50 response to the first stimulus in normal subjects is typically reasonably large, but the P50 response to the second stimulus is attenuated by inhibitory effects associated with the first stimulus (37,73). This temporary inhibitory effect presumably serves to protect processing of the initial stimulus from the potentially disruptive effects of the second stimulus. Schizophrenic subjects, in contrast, often do not show this decreased P50 response to the second stimulus (1). While about 90% of control subjects without a family history of schizophrenia show a reduction of more than 50% in the amplitude of the second stimulus relative to the first, about 85% of schizophrenic patients show less than a 50% reduction (37). This inhibitory deficit is seen in both unmedicated and medicated patients (33). Independent groups have been able to replicate this basic P50 inhibitory deficit in schizophrenia (11, 65).

Within a small sample of schizophrenic patients, the extent of the P50 gating deficit has been found to correlate with the level of performance deficit in sustaining focused attention (19), suggesting that the P50 gating deficit may be related to one of the most prominent enduring neurocognitive performance deficits in schizophrenic patients (82). Cullum et al. (19) suggest that the P50 amplitude to the first stimulus might reflect the normal process of involuntarily attending to an initial relatively rare event, while the P50 to the second stimulus may indicate the extent to which involuntary attention is again drawn to a second example of an unimportant stimulus. Thus, although the P50 sensory gating abnormality is often assumed to reflect relatively automatic processes, the extent to which various aspects of attention are involved in this phenomenon is of substantial interest (118).

The abnormality in P50 gating is also present in a disproportionate number of the first-degree biological relatives of schizophrenic patients (101,116). For biological relatives, the P50 gating deficit has been associated with increased Sc scores on the Minnesota Multiphasic Personality Inventory. In a sample of nine extended families with multiple cases of schizophrenia, Freedman et al. (35) recently found that the P50 gating deficit was linked to the site of the 7 nicotinic receptor on chromosome 15. This linkage is consistent with evidence (discussed below) suggesting that this neuronal nicotinic receptor may act in an inhibitory pathway that influences the P50 response to the second stimulus. While replication of this linkage is clearly needed, this finding serves as an excellent example of the use of a schizophrenia-related endophenotype to search for inherited components of vulnerability to schizophrenia.

Another major thrust of P50 research has been directed at specifying components of a neuropsychopharmacological model of this sensory gating abnormality. As typical neuroleptic medications do not normalize the P50 gating deficit, excessive dopaminergic activity is unlikely to account for this phenomenon. An initial study suggests that clozapine, on the other hand, can normalize the P50 sensory gating abnormality in clinically responsive patients (73). Freedman, Adler, Waldo and their colleagues (33,34) have demonstrated that the P50 gating phenomenon may involve nicotinic cholinergic mechanisms at the level of the hippocampus and temporal lobes and have emphasized hippocampal mediation of the gating deficit. An animal model of the sensory gating mechanism, which uses a P20/N40 auditory response rather than P50 to index gating, has shown that hippocampal interneuron response is strikingly suppressed to the second stimulus, relative to the first (34). Other results suggest that cholinergic neurons in the septal region also show gating of auditory stimuli (73).

The role of the 7 nicotinic acetylcholine receptor has been a particular focus of recent pharmacological work on the P50 gating effect. The P50 inhibitory deficit in schizophrenic patients and in their unaffected biological relatives appears to be transiently normalized by nicotine (2,73). Inhibitory gating of response to the second auditory stimulus in the animal model has been found to be reduced specifically by antagonists of the low-affinity nicotinic receptor, 7 (73). Leonard et al. (73) presented a neuronal model of auditory gating in the hippocampus that involves interneuron inhibition of a pyramidal cell through GABAergic synapses, and stimulation of the interneuron and the pyramidal cell by cholinergic input from the septum. Blocking the septal cholinergic input removes the inhibitory effect of the interneuron and allows the pyramidal cell to fire in response to the second auditory stimulus. Additional work is needed to determine the degree to which the neural mechanisms that account for the P50 gating phenomenon overlap with those involved in another popular procedure for examining sensorimotor gating—prepulse inhibition of the startle reflex response.

The "prepulse inhibition" (PPI) paradigm for study of sensorimotor gating also involves the presentation of two stimuli in close succession, but it examines the inhibition of the amplitude of the startle reflex rather than the attenuation of the P50 response. Operationally, it consists of presenting a startle-eliciting stimulus pulse (e.g., a burst of loud noise) sometimes alone and sometimes shortly following a non-startling stimulus (e.g., a mild tone). As shown in Figure 3, when the non-startling "prepulse" precedes the startling "pulse" by an interval ranging from approximately 30–300 ms, the startle reflex is markedly inhibited compared with when the startle pulse is presented alone. On average, the startle reflex amplitude is inhibited by 50% or more, and in some individuals it is completely suppressed. The PPI phenomenon occurs reliably in lower animals using measures of whole-body startle (55), as well as in humans using the eye-blink component of the startle reflex (4,44). The optimal interval between onset of the prepulse and onset of the startle pulse for producing PPI in both humans and animals is approximately 100 ms.

In human research, PPI is usually measured as a change in eye-blink amplitude, because the eye-blink is one of the most reliable components of the startle reflex and is relatively easy to measure with electromyographic techniques. The startle-eliciting pulse is usually a brief burst of loud noise (95–105 dB) with a rapid rise time, although tactile stimuli (air puffs) and visual stimuli (bright flashes) also have been used effectively. The non-startling prepulse is usually a brief mild innocuous tone, but it, too, can be in other modalities. The prepulse may be discrete (onset and offset before the startle pulse is presented) or it may be continuous (sustained until or beyond the presentation of the startle pulse).

Graham (44) suggested that the PPI may reflect an automatic sensory gating mechanism initiated by the prepulse that protects initial processing of the prepulse from the distractive effects of other sensory events, such as startle stimuli. That is, the onset of a low-intensity prepulse triggers an inhibitory mechanism that momentarily attenuates reactions to extraneous stimulation while the prepulse receives early perceptual processing. In support of this view are recent reports that the prepulse is more accurately perceived in trials on which PPI occurs (81) and in individuals who exhibit the greatest PPI (30). Further research beyond these preliminary studies is needed to confirm the functional significance of PPI.

Although PPI was thought initially to be purely automatic and to be completely independent of attention effects, early results suggested that attentional factors could influence the degree of PPI (28, 49). More recent studies have even more convincingly demonstrated that directing attention to the prepulse enhances the PPI effect (e.g., 31). In the latter "attention to prepulse" paradigm, normal college students were presented with an intermixed series of two types of prepulses, high-pitched and low-pitched tones, and were instructed to attend to one type and ignore the other. The attended and ignored prepulses were then followed by an acoustic startle stimulus at lead intervals of 60, 120, and 240 ms. Significantly greater PPI of the startle eyeblink reflex was found at the 120-ms lead interval following the onset of the attended prepulse, compared with the ignored prepulse. The attention effect was not present at the 60 or 240 ms lead intervals. Thus, attention directed to a prepulse has a time-locked modulatory influence on PPI.

The demonstration of attentional modulation of PPI is important for at least two reasons. First, it indicates that PPI is not solely an automatic process in all situations, because it can be modulated by conscious attentional strategies. The fact that PPI occurs at 60 ms without attentional modulation, and at 120 ms with attentional modulation, suggests that the measurement of PPI within the "attention to prepulse" paradigm allows evaluation of both automatic and controlled cognitive processing within one paradigm (27). Second, as attentional deficits have long been recognized as a key feature of schizophrenia (84), the attentional modulation of the PPI effect may have utility as a sensitive index of schizophrenic impairments attributable to faulty use of selective attention.

Several studies have examined PPI in heterogeneous groups of hospitalized chronic and acute medicated schizophrenic patients (14,15,45). Braff et al. (15) employed a continuous mild tone as the prepulse and a burst of loud white noise as the startle pulse. Braff et al. (14) employed a discrete mild white noise as the prepulse and either a burst of loud white noise or a tactile stimulus as the startle pulse. Grillon et al. (45) used a discrete white noise as the prepulse and a burst of loud white noise as the startle pulse. In each study, schizophrenic patients exhibited impaired PPI, compared with normal control subjects when the prepulse preceded the startle stimulus by between 60 and 120 ms. Impaired PPI was documented across different modalities of the startle stimulus (14) and across different intensities of the prepulse (45). These findings have been interpreted as reflecting an impairment in automatic, preattentive central nervous system inhibition (sensorimotor gating) in schizophrenia (13). The impaired sensorimotor gating may underlie the vulnerability in schizophrenia to sensory flooding, cognitive fragmentation, and conceptual disorganization (12, 13).

Dawson et al. (21) extended this line of research by studying attentional modulation of PPI in relatively asymptomatic schizophrenic outpatients. In contrast to the previous studies, in which patients were not given specific instructions to attend to the stimuli. The "attention to prepulse" paradigm was used, in which subjects are instructed to attend to one type of auditory prepulse (e.g., a high pitched-tone) and to ignore another type (e.g., a low-pitched tone). Normal subjects demographically matched to the schizophrenic patients exhibited greater PPI following the attended prepulse than the ignored prepulse at the 120 ms lead interval, but not at the 60 ms lead interval. In contrast, the patients failed to exhibit differential PPI following the attended and ignored prepulses at any lead interval. The PPI of the schizophrenic patients differed from that of the normal subjects (Figure 4) by failing to show enhanced inhibition at 120 ms following the attended prepulse; the PPI of the patients and controls did not differ at 60 ms following either the attended or ignored prepulse. These findings indicated defective attentional modulation of PPI by schizophrenic outpatients, suggesting impaired inhibitory protection of the processing of significant stimuli. The fact that these findings were obtained for relatively asymptomatic outpatients suggested that the failure to attentionally modulate PPI might be a trait-like index related to ongoing vulnerability to schizophrenia, rather than a secondary effect of psychotic symptoms.

Preliminary results examining PPI in both passive and active attentional paradigms within the same schizophrenic patients are beginning to be reported (77). McDowd et al. (77) measured PPI from small groups of "late-life" (mean age = 55 years) schizophrenic patients and nonschizophrenic control subjects (mean age = 66 years) in both passive and active paradigms. During the first, passive, phase of the experiment, subjects were told to sit quietly, whereas in the second, active, phase they were instructed to listen carefully for the prepulse and to press a button each time it was detected. The schizophrenic subjects showed less PPI than the control group, particularly in the active attention phase. Unfortunately, this paradigm did not allow separate evaluation of the effects of "arousal" versus "attention," because the passive and active tasks were given in separate phases. Changes in tonic arousal level over time or in generalized activation as a result of the response requirement are not separable from specific effects of the attentional instructions. An important direction for future research is to assess more clearly the passive and active attentional effects on PPI within the same group of patients.

One of the methodological advantages of the PPI phenomenon is that it can be studied in both humans and animals (e.g., 110). The animal studies have added significantly to our understanding of the neurobiological mechanisms underlying PPI and its impairments. For example, a neural circuit for the primary acoustic startle reflex has been established in rats. Research in the early 1980s (20) suggested that the acoustic startle circuit consisted of the auditory nerve, ventral cochlear nucleus, nuclei of the lateral lemniscus, nucleus reticularis pontis caudalis, spinal neuron, and muscle. More recent research (71, 75, 122) indicates an even simpler circuit that involves axons from the cochlear root neurons in the auditory nerve projecting directly to the nucleus reticularis pontis caudalis, and then to the spinal neurons and muscles.

More relevant to the present discussion of PPI, neural circuits are being established for the modification of the startle reflex. Leitner and his co-workers have proposed a basic midbrain circuit for PPI by auditory prepulses based on brain lesion findings in rats (e.g., see ref. 72). This hypothesized circuit for auditory prepulse input consists of the auditory nerve, ventral cochlear nucleus, nuclei of the lateral lemniscus, the inferior colliculus and lateral tegmentum, from the point where it then descends to the hindbrain "startle center" of the caudal pontine reticular nucleus. Cells in the pedunculopontine nucleus of the lateral tegmentum send inhibitory signals to the reticularis pontis caudalis (68), resulting in the inhibition of the startle reflex elicited shortly following onset of the prepulse. Leitner and Cohen (72) also suggest that the lateral tegmental area serves as a convergence point for descending information from prepulses in other sensory modalities.

Although the primary startle reflex circuit is located in the hindbrain, and the basic PPI circuit appears to be located in the midbrain, there is considerable evidence that forebrain structures can modulate PPI (42, 110, 114). This, of course, would be expected, given that higher order attention can modulate PPI. PPI can be modulated by a complex cortico-striato-pallidal-pontine circuit. Based on the research of these investigators and others, Figure 5 shows a proposed modulatory circuit including innervation from the medial prefrontal cortex and hippocampus to the nucleus accumbens in the striatum, striatal connections to the pallidum, and pallidal input to the pedunculopontine nucleus of the tegmentum, with the final inhibitory input from the pontine tegmentum to the nucleus reticularis pontis caudalis (see ref. 27 for an overview of this circuit; see refs. 110 and 111 for more details).

PPI is disrupted by systemic administration of dopamine agonists such as apomorphine or glutamate antagonists such as PCP (110). In addition, pharmacological manipulations in the proposed modulatory circuit have disruptive effects on PPI. For example, PPI is reduced by hippocampal infusion of the cholinergic agonist carbachol (17), NMDA infusion in the ventral subiculum (117), dopamine infusion within the nucleus accumbens (112), and by GABAergic activity in the ventral pallidum (109). The striatal dopaminergic modulation of PPI involves primarily D2, and not D1, receptors (see review in ref. 111). Thus, the disruptive effects of the dopamine agonist apomorphine are blocked by the D2 antagonists raclopride and spiperone, but not by the D1 antagonist SCH 23390 (113). Moreover, the lability of antipsychotics to restore PPI in apomorphine-treated rats is highly correlated with their clinical efficacy (110). Data such as these support the suggestion that impaired PPI in rats may be a valid animal model of deficient sensorimotor gating in schizophrenic patients (110) and for testing antipsychotic activity of novel compounds (108).

The study of PPI is currently enjoying an upsurge of interest in both human and animal research. We believe that we are entering a period of heightened research activity on this phenomenon that will focus on a number of clinical and preclinical issues in the next few years. One of the more prominent clinical issues is whether the PPI deficits are state-related episode indicators or trait-related vulnerability indicators in schizophrenia and other disorders. The relation of the PPI deficits to symptom dimensions, subtypes, course, and prognosis of schizophrenia is another critical area in need of research. Evidence of a relationship between PPI deficits and an index of thought disorder (the Ego Impairment Index) is one example of this direction (88). The effects of various drugs on PPI will also continue to be an active area of research linking the preclinical and clinical domains.

Among the more important preclinical issues are a better definition of the neural substrates of PPI and its modulation and their relationships to psychobiological deficits in schizophrenia. State-of-the-art neuroimaging techniques will no doubt be used in conjunction with the PPI paradigm to productively address these issues. For example, results of PET scans with fluorodeoxyglucose (FDG) are beginning to be reported with the "attention to prepulse" paradigm in unmedicated schizophrenic patients and controls (50). The patients failed to show attentional modulation of PPI, consistent with earlier findings (21) and had significantly lower glucose metabolism in the frontal lobes during the task. Greater understanding of PPI and its attentional modulation will likely emerge from the use of newer brain imaging techniques with much better temporal resolution, such as functional magnetic resonance imaging. Use of topographic mapping with ERP or magnetoencephalographic procedures (see Molecular Analys of the Single Cell: Importance) combined with PPI paradigms might allow even finer temporal resolution for examining brain circuits that mediate PPI and its attentional modulation in the first hundred milliseconds after stimulus onset.

The significance of PPI within an information processing theoretical framework will also be a focus of important future research. The distinction between the passive PPI phenomenon and attentional modulation of PPI may be a critical issue. The time course of PPI may be a revealing measure here, because passive PPI can occur at very short lead intervals (30–60 ms), whereas attentional modulation is evident only at longer lead intervals (120 ms). The differential time course suggests different underlying mechanisms, and it will be interesting to determine whether both mechanisms have the same relationship to schizophrenia. PPI appears to have both involuntary automatic and controlled voluntary components, and the "attention to prepulse" paradigm may offer one way to parse the deficits in these two types of processes in schizophrenic patients (12). The multiple neuroanatomical and neuropharmacological modulatory effects on PPI have been demonstrated almost exclusively within the passive PPI paradigm; it remains to be seen how these effects differ under more active, attention-demanding conditions. In summary, the study of PPI is an excellent example of a model system through which multidisciplinary research can help to integrate cognitive science, clinical science, and neuroscience (27).

Electrodermal activity (EDA) refers to several aspects of electrical phenomena measured from the skin, particularly changes in the electrical conductivity of the skin. EDA is usually measured by passing a small, imperceptible electrical current through a pair of electrodes on the surface of the skin, typically on the volar surfaces of the finger tips or on palmar sites. One can then measure the electrical conductivity of the skin and the changes in that conductivity that occur under different stimulus conditions. EDA is a popular measure in human psychophysiology because of its sensitivity to psychological states and processes, combined with its relative ease of measurement and quantification. The reader can refer to Boucsein (10) and Roy et al. (93) for book-length treatments of EDA and to Dawson, Schell, and Filion (26) and Hugdahl (61) for shorter, chapter-length reviews. A brief overview of EDA is available in Dawson and Nuechterlein (22).

The principal peripheral effectors mediating skin conductance are the eccrine sweat glands, which are quite dense on the palmar surfaces of the hands. Increases in sweat gland activity, which lead to increases in skin conductance, are under the neural control of the sympathetic branch of the autonomic nervous system. Therefore, EDA is a noninvasive peripheral index of sympathetic arousal and reactivity.

Electrodermal activity can be divided into tonic and phasic measures. The phasic skin conductance response (SCR) is a momentary increase in skin conductance elicited with a latency of 1–3 sec following onset of a novel, unexpected, significant, or aversive stimulus. If the eliciting stimulus is innocuous, the SCR is considered to be a component of the "orienting response," and it will habituate after several presentations of the stimulus. The phasic SCR is thought to reflect the subject's attention to, and cognitive processing of, the eliciting stimulus (100). Phasic SCRs also can occur in the absence of specific identifiable eliciting stimuli and are called nonspecific skin conductance responses (NS-SCRs).

Tonic electrodermal measures are stable or slowly changing aspects of skin conductance. The most common tonic EDA measure is skin conductance level (SCL). SCL will typically be relatively high in the novel laboratory environment, then will decrease gradually while subjects are at rest, and increase when stimulation is introduced or some task is required. Both SCL and NS-SCRs are considered useful indices of sympathetic arousal, with high SCL and more frequent NS-SCRs indicating higher sympathetic arousal.

EDA research with schizophrenia patients has focused primarily on phasic SCR orienting responses to innocuous stimuli (e.g., moderately intense tones without task significance). Approximately 40%–50% of schizophrenic patients fail to exhibit any SCR orienting responses to innocuous tones, compared with only 5%–10% in the normal population (7,22,123). In contrast to the SCR "non-responders", the remaining 50%–60% of the patients do respond with SCRs to innocuous tones but also show abnormally high tonic arousal. Thus, there appear to be at least two well documented EDA abnormalities in schizophrenia: (1) SCR hyporesponsiveness to innocuous stimuli and (2) SCL and NS-SCR hyperarousal (22,23).

A key issue in current research concerns the meaning of the two types of EDA abnormalities found in schizophrenic patients. This issue has been addressed by relating phasic and tonic EDA measures to symptoms and prognosis and by determining whether the EDA measures change with alterations in symptomatic state when both are assessed longitudinally.

Regarding EDA relationships with current symptoms, several early studies suggested that SCR non-responding tends to be related to presence of negative symptoms (8,105). Others, however, have suggested that the relationship may not be this simple. For example, Bernstein (6) later suggested that SCR non-responding may be related specifically to a subgroup that is conceptually disorganized and emotionally withdrawn, thereby displaying both positive and negative symptom features. SCR hyporesponsivity has been related to several indices of more severe illness (66) and poor premorbid adjustment (86), although phasic and tonic hyperactivation have also been found to be associated with higher levels of overall symptomatology in a group of young recent-onset schizophrenic patients (23).

Regarding relationships between EDA and prognosis, several studies have reported that high EDA responsivity or activation measured during the inpatient phase is associated with poor short-term symptomatic outcome (23,41, 106,124). However, SCR non-responding, rather than hyperactivation, has been associated with poor social and employment outcome in men (86), but not in women (119), and with shorter time to relapse following hospital discharge (62). Factors that might account for these apparently discrepant results (e.g., gender of patient, acute versus chronic status of patient, definition of outcome) have not yet been identified.

Longitudinal studies of EDA and symptomatic changes are now beginning to appear. Dawson et al. (24), for example, recorded EDA from schizophrenia patients and matched normal controls on two occasions—one while the patients were in a remitted state and the other while the same patients were in a psychotic state. The primary result was that tonic EDA became abnormally high only during the psychotic episode. Phasic SCR hyporesponsiveness to innocuous stimuli, when corrected for the overall arousal level, was present in both the remitted and psychotic states. These findings suggest that schizophrenia psychotic episodes are frequently accompanied by high sympathetic nervous activation. Moreover, longitudinal data from five remitted schizophrenic patients suggested that increased EDA activity might occur in the week or two before a psychotic exacerbation (53). These findings are consistent with a theoretical framework which hypothesizes that sympathetic activation is a "transient intermediate state" that precedes psychotic episodes in vulnerable individuals (83, 85). Thus, tonic electrodermal measures might be useful in understanding the nature of the process by which psychotic symptoms return and in obtaining objective early warning signs of that process.

In contrast to the longitudinal results described above, some other studies have reported that both phasic and tonic EDA were relatively stable across different symptomatic states (87, 102, 124). The previous studies differed in several ways from that of Dawson et al. (24), and it is clear that more longitudinal research is needed to fully test the state versus trait role of EDA in schizophrenia.

Another interesting issue concerns the relationship of EDA to genetic vulnerability factors for schizophrenia. Adolescents at high genetic risk (having two parents with a schizophrenia-spectrum diagnosis) showed impaired habituation of tonic EDA arousal over a 15-minute rest period; those with one spectrum parent exhibited some habitation; and those with normal parents evidenced rapid habituation (58). These data suggest that impaired habituation of sympathetic nervous system arousal may represent an indicator of the genetic predisposition to schizophrenia. Moreover, adolescents at genetic risk who had high EDA arousal and had experienced severe disruption of the early family rearing environment may be particularly vulnerable to positive symptom schizophrenia, whereas those who are SCR nonresponders and had suffered delivery complications may be particularly vulnerable to negative symptom schizophrenia (18).

The central neurophysiological and pharmacological mechanisms mediating EDA appear to be multiple and complex (10, 93, 98). Bouscein (10) describes at least two and possibly three independent cerebral pathways for EDA. The first involves hypothalamic and limbic structures, with excitatory influences stemming primarily from the amygdala and inhibitory influences stemming mainly from the hippocampus. The second involves the basal ganglia together with the premotor cortex. The third involves the reticular system, which may have eliciting as well as modulating influences on EDA.

The neurochemical bases of EDA also appear to be complex and are not well understood at this time. In rats and cats, Yamamoto and colleagues (120,121) found that chemical lesions in the ascending noradrenergic fibers completely abolished phasic SCRs and also reduced tonic SCL and NS-SCRs for several days. Several weeks later, most of the rats showed a recovery of the tonic EDA and a return of the phasic SCR that did not habituate. Destruction of the dopamine system did not affect either phasic or tonic EDA. These data are consistent with those from other animal studies that suggest an important role for norepinephrine, but not for dopamine, in the control of EDA. In contrast, as summarized by Zahn and colleagues (125), human pharmacological studies suggest that dopamine activity is positively related to EDA. In support of the latter conclusion, Zahn et al. (125) found that the metabolites of dopamine and serotonin, but not of norepinephrine, were positive correlated with phasic SCRs in a group of subjects with obsessive-compulsive disorder. The authors concluded that these data "support hypotheses of the influence of the nigrostriatal dopamine system on the generation of EDA" (125).

Brain imaging techniques promise to enhance greatly our understanding of the central origins of EDA. For example, the first preliminary PET study of three SCR responder and three non-responder schizophrenia patients has been reported (51). In comparison to both normal controls and responder patients, the non-responders showed about a 20% reduction in metabolic rate across the entire brain. Non-responders also had significantly lower relative metabolic rates in medial frontal and hippocampal areas, as well as in the right amygdala, compared with responders. More studies are needed, with larger samples of schizophrenic patients and also samples of SCR responders and non-responders from the normal population, to determine the reliability and specificity of these preliminary PET/EDA relationships (see review in ref. 90).

A potentially important aspect of EDA measures in schizophrenia is that they reveal two distinct abnormalities in different subgroups: (1) SCR hyporesponsiveness to innocuous environmental stimuli and (2) SCL and NS-SCR hyperarousal. The search for the meaning of these two abnormalities has focused primarily on symptomatic, prognostic, and genetic risk correlates.

A major direction for future research is to also determine the cognitive and the neurophysiological meaning of the EDA abnormalities. At the cognitive level, for example, SCR orienting hyporesponsiveness may be related to inefficient allocation of attentional resources to environmental stimuli, which may index a vulnerability factor for schizophrenia (52). At the neurophysiological level, the use of neuropsychological testing and brain imaging may help to delineate the meaning of the EDA abnormalities (46,90). The EDA measures should contribute further to the study of schizophrenia when the cognitive and neurophysiological significance of the individual differences in tonic and phasic EDA are better understood.

A small change in an environmental stimulus can trigger a cascade of neurophysiological and psychophysiological responses, some of which occur in parallel and others in serial fashion. The measures reviewed in the present chapter (P50 and P300 event-related potential components, prepulse inhibition of startle blink, and the skin conductance response) reflect some of the events and processes that occur within the first few hundred milliseconds following stimulus onset. Despite the marked differences in the specific neurophysiological and psychophysiological processes being examined, all of these measures detect abnormalities in schizophrenia. At a minimum, these abnormalities indicate that schizophrenia is characterized by impairments in early components of sensation, perception and information processing following stimulus onset. Some of these abnormalities may also reflect genetic factors conveying susceptibility to schizophrenia.

One of the earliest processes following stimulus onset involves the automatic detection of a stimulus change, whereas later processes have to do with the recognition and interpretation of the meaning of the change. The detection of a stimulus change is normally associated with sensory or sensorimotor gating mechanisms that serve to protect further processing of the detected stimulus from interference by stimuli that immediately follow. Two of the measures reviewed here (P50 suppression and PPI) are believed to index early sensory or sensorimotor gating. Both PPI and P50 suppression may serve as objective measures of the ability of one stimulus to attenuate reactions to immediately following stimuli. The evidence of impaired PPI and P50 suppression is consistent with clinical observations and laboratory performance data which suggest that schizophrenic patients have deficient gating and are less able to screen out irrelevant, unwanted stimuli (12, 78, 84). However, it should be noted that when measured in the same subjects, there is only a weak relationship between the amount of P50 suppression and the amount of PPI (97). This weak correlation suggests that PPI and P50 suppression reflect largely different cognitive and neural processes, despite broad conceptual similarities. Additional research is needed to determine the extent to which factors such as the clinical state of the patient, attentional processes, and emotional state affect PPI and P50 suppression in schizophrenia in a distinctive fashion.

In contrast to PPI and P50 suppression, the P300 ERP component and the SCR are both thought to reflect controlled (or effortful) cognitive processing of the eliciting stimuli that occur after stimulus onset has been detected. Both P300 and SCRs are hypothesized to index orienting and attentional allocation and both are normally very responsive to the task relevance of the stimulus. Again, it is noteworthy that schizophrenic patients generally exhibit smaller than normal P300s and SCRs, thereby providing converging evidence of impairments in controlled processing of environmental stimuli in schizophrenia. Deficits in controlled or effortful information processing have been a prominent feature of several accounts of schizophrenic neurocognitive performance deficits (e.g., 16, 84), an emphasis that has most recently focused on working memory abnormalities. However, P300 and SCR habituate at different rates and are differentially affected by various independent variables. Thus, the SCR and P300 measures do not convey identical information, but rather index somewhat different aspects of information processing mechanisms. Furthermore, given what is known about the brain structures involved in generation of SCR and P300, it appears that they reflect different neural substrates.

Thus, schizophrenic patients exhibit abnormalities in several neurophysiological and psychophysiological processes that occur shortly following stimulus onset. Some of these abnormalities, such as those indexed by P50 suppression and P300 latency, are also present among a disproportionate number of first-degree relatives of schizophrenic patients, suggesting that they may index genetic vulnerability factors. Impaired habituation of tonic electrodermal activation over time may also reflect a dimension of genetic predisposition. In addition to serving as tools for understanding the underlying nature of schizophrenia and related genetic predispositional factors, certain neurophysiological and psychophysiological measures may also greatly aid treatment development. Animal paradigms for measuring abnormalities that parallel human PPI and P50 suppression, for example, offer new models for evaluating new of antipsychotic agents.

A more complete conceptual integration of the different neurophysiological and psychophysiological abnormalities characterizing schizophrenia is certainly desirable, but is hampered at this point by the lack of relevant data. Because these abnormalities have typically been examined in separate laboratories with different subjects, the empirical relationships among these anomalies are relatively unexplored. Further systematic research is also needed to delineate further the relationships of the individual neurophysiological and psychophysiological abnormalities to specific cognitive and neural processes. In addition, research is urgently needed to determine whether these abnormalities reflect different or related underlying dimensions of vulnerability to schizophrenia. Finally, as schizophrenic patients vary greatly in the extent to which they show these neurophysiological and psychophysiological anomalies, the utility of such basic abnormalities to shed light on the diverse course of schizophrenia and on differential response to treatment deserves much more attention.

published 2000