Eyeblink conditioning deficits indicate timing and cerebellar abnormalities in schizophrenia

Abstract

Accumulating evidence indicates that individuals with schizophrenia manifest abnormalities in structures (cerebellum and basal ganglia) and neurotransmitter systems (dopamine) linked to internal-timing processes. A single-cue tone delay eyeblink conditioning paradigm comprised of 100 learning and 50 extinction trials was used to examine cerebellar timing circuits in 13 medicated patients with schizophrenia and 13 age- and sex-matched controls. Patients with schizophrenia showed impaired learning of the conditioned response compared to controls and also greater within-subject variability in the timing of their responses. These findings are consistent with models of schizophrenia in which timing deficits underlie information-processing abnormalities and clinical features of the disorder.

Introduction

Intra-individual response variability and behavioral heterogeneity were recognized as essential features of schizophrenia (SZ) in the influential nosologies of Bleuler (1911/1950) and Kraepelin, Barclay, and Robertson (1919). In fact, classic symptoms of SZ such as formal thought disorder, disorganized and bizarre behavior, and soft neurological signs can be viewed as manifestations of a disturbance of temporal coordination of information processing in the central nervous system. Some contemporary theoretical models describe intra-subject response variability as a “fundamental characteristic” (Lehmann & Cancro, 1985) and qualitative feature of SZ (Meehl, 1990). Not surprisingly, short-interval timing deficits in SZ have been reported using a variety of methods, including time estimation as well as temporal production and reproduction tasks (Densen, 1977, Johnson and Petzel, 1971, Tracy et al., 1998, Tysk, 1983, Tysk, 1990, Volz et al., 2001). In addition, reaction time (e.g., Manoach et al., 2001, Shakow, 1962) and event-related brain potential studies (e.g., Ford et al., 1994, Matthysse et al., 1999, Patterson et al., 2000) report greater within-subject temporal response variability among individuals with SZ compared to non-patient comparison groups. The purpose of the present investigation is to use eyeblink conditioning methodology to examine the integrity of cerebellar and related neural-timing circuits in SZ. The impetus for the present research arises from models of SZ in which within-subject timing deficits underlie information-processing abnormalities and clinical features of the disorder.

Although relatively little theoretical consideration has been given in the literature to within-subject response variability in SZ, a hand-full of models of SZ assert that increased intra-individual response variability is an underlying feature of the illness and specifically predict variable response timing. For example, Meehl, 1962, Meehl, 1989, Meehl, 1990 postulated that the fundamental problem in SZ is a ubiquitous neural integrative defect, termed schixotaxia, that is associated with aberrant post-synaptic firing probabilities. This deficit is said to produce psychophysiological and soft neurological aberrations and forms of cognitive “slippage,” such as loose associations and thought disorder (Meehl, 1989, Meehl, 1990). Meehl (1990) reasoned that this “molar slippage” in cognitive and cognitive-affective processes was indicative of “microslippage” at the neuronal level. Accordingly, the model predicts that “microslippage,” via aberrations in the synaptic control of neuronal firing probabilities, introduces intra-individual temporal response variability in SZ.

Another influential model to emerge from the cognitive information-processing revolution of the 1950s and 1960s was the segmental set theory proposed by Shakow, 1962, Shakow, 1963. Shakow postulated that perceptual, integrating, and organizing functions of the brain, which underlie the establishment of generalized states of readiness for responding to incoming stimuli, are impaired in SZ. In the healthy individual, major “sets” facilitate and economize information processing and responding by providing a parsimonious and consistent strategy for processing and responding to stimuli in the environment. Shakow proposed that individuals with SZ have difficulty maintaining the optimal, situationally determined “state of readiness” (Nopoulos, Ceilly, Gailis, & Andreasen, 2001, p. 9) for responding to external and internal inputs due to susceptibility to distraction by irrelevant aspects of the stimulus environment (both internal and external). This inability to maintain the contextually relevant set is secondary to a tendency to form multiple, disarticulated, segmented sets (or minor sets) that are elicited by transient situational irrelevancies. Temporal response variability, then, occurs because successive adoption of minor, segmented sets disrupts and precludes the fluid processing of information through the central nervous system.

In a similar and more recent theoretical formulation, Matthysse et al. (1999) proposed that SZ is associated with intermittent lapses or degradation of performance, termed dialipsis, which is caused by impaired cognitive control mechanisms that modulate the effect of motives on cognition and perception. According to the authors, “The intermittent degradation model accounts for the higher variance within individuals in the schizophrenic group by the erratic presence of the dialipsis component, and for the variability of the variance from person to person within that group” (p. 20). The authors postulate that not all individuals with SZ are susceptible to dialipsis. As with Meehl and Shakow, Matthysse and colleagues proposed that the organismic instability of perceptual and cognitive processes, which characterizes dialipsis in affected patients, underlies core features of SZ such as thought disorder.

“Disconnection” models of SZ (Friston, 1998) offer additional ways to understand the possible role of timing problems in SZ. One of the more prominent of such models, Andreasen’s (1999) “cognitive dysmetria” model, states that the symptoms of SZ indicate a disturbance in the fluid temporal coordination of motor, perceptual, and cognitive sequences of behavior. She hypothesized that this synchronous, temporally coordinated organization of behavior is controlled by a neural circuit comprised of cortico-cerebellar-thalamic-cortical (CCTC) connections. Interestingly, Andreasen proposed that the heterogeneity of the symptoms of SZ, including thought disorder, disorganized behavior, and soft neurological signs, may be unified by postulating a basic cognitive deficit that results from abnormalities in the CCTC circuit.

Taken together, these theoretical models provide strong impetus for examining the functional integrity of neural timing mechanisms in SZ. However, there is generally a paucity of empirical research linking timing deficits in SZ to specific neural circuits and/or mechanisms. Exceptions can be found in the work of Andreasen and her colleagues (Andreasen, Paradiso, & O’Leary, 1998) and others (Buhusi and Meck, 2002, Meck, 1996, Rammsayer, 1997, Tysk, 1983, Tysk, 1990, Volz et al., 2001), which in various combinations relate timing deficits and structural brain abnormalities or neurotransmitter systems to SZ.

There is accumulating evidence that brain structures linked to response timing (e.g., cerebellum and basal ganglia) are abnormal in SZ. For example, deficits in the cortical and thalamic nodes of the CCTC circuit have been clearly demonstrated (Nopoulos et al., 1999, Staal et al., 2001, Volz et al., 2000), in addition to abnormalities in the cerebellar node (Andreasen et al., 1996, Andreasen et al., 1997, Vrtunski et al., 1996), which has been widely linked to response timing (e.g., Fiala et al., 1996, Ivry and Keele, 1989, Spencer et al., 2003, Steinmetz, 2000; see below). In addition, decreased cerebellar size has been observed in SZ (e.g., Ichimiya et al., 2001, Lippman et al., 1982, Loeber et al., 2001, Nopoulos et al., 1999, Weinberger et al., 1979; but see negative findings from Nasrallah, McCalley-Whitters, & Jacoby, 1982) and is a reliable indicator of poor long-term outcome (Wassink, Andreasen, Nopoulos, & Faum, 1999). A correlation between reduced cerebellar vermal volume and total BPRS Depression and Paranoia subscore has also been reported (Ichimiya et al., 2001). Moreover, the observed cerebellar volume deficits are correlated with greater cognitive dysfunction (Nopoulos et al., 1999), lending support to the theory that cerebellar dysfunction contributes to “cognitive dysmetria” (Andreasen, 1999). Additionally, correlations between abnormal morphologies of multiple structures implicated in the CCTC circuit have been reported. Cerebellar anterior vermis volume has been positively correlated to temporal lobe volume (Nopoulos et al., 1999), and midbrain and cerebellar vermis size were also positively correlated in SZ patients (Nopoulos et al., 2001). Finally, the potential importance of the CCTC circuit in SZ is further underscored by the fact that feedback and feedforward loops are widely known to connect the cerebellum with areas of the brain implicated in the disorder. For example, the cerebellum is especially well connected with the thalamus and limbic system (Anand et al., 1959, Snider et al., 1976), and prefrontal cortex (Schmahmann & Pandya, 1995). Taken together, these morphometric and metabolic findings, and the connectivity of the cerebellum with brain areas implicated in SZ strongly suggest a role for cerebellar dysfunction in the disorder.

Importantly, cerebellar dysfunction in SZ may contribute to timing problems given compelling evidence that the cerebellum plays a fundamental role in the timing of neural processes associated with perceptual and cognitive functioning (Ivry and Keele, 1989, Katz and Steinmetz, 2002, Leiner et al., 1991). Patients with a variety of cerebellar lesions have demonstrated deficits in both time perception and production tasks (Ivry & Keele, 1989), and display increased variability during rhythmic tapping tasks, which has been linked to internal timing mechanisms independent of motor dysfunctions (Ivry, Keele, & Diener, 1988). Furthermore, increased regional cerebral blood flow (rCBF) in both cerebral hemispheres and the inferior vermis has been demonstrated using positron emission tomography (PET) during an auditory interval discrimination task (Jeuptner et al., 1995). Moreover, patients with localized cerebellar lesions manifest behavioral and cognitive disturbances that are remarkably similar to those seen in SZ, including impairment of executive functions, difficulties with visual–spatial organization and memory, blunted affect and inappropriate behavior, and language deficits (Schmahmann & Sherman, 1998).

The basal ganglia, another structure implicated in response timing, has also been shown to be aberrant in SZ. Both overall volume reduction and imbalanced activity between the right and left basal ganglia have been documented in SZ (Seeman, 1993), as well as reduced absolute metabolic rates in patients as compared to controls. Furthermore, the basal ganglia maintain connections to neural structures previously implicated in the disorder including the frontal lobe and the thalamus. Given the goal-directed behavior and action planning functions of the basal ganglia, it would be reasonable to propose that deficits in this structure could explain a variety of the symptoms of SZ such as motor stereotypies, disordered cognitive planning, and affective apathy (Graybiel, 1997). Basal ganglia dysfunction may also contribute to the timing deficits characteristic of SZ because the role of the basal ganglia in interval timing has been well documented. For example, an event-related fMRI study reported activation of the basal ganglia when encoding time intervals (Rao, Mayer, & Harrington, 2001), and a series of animal lesion and ensemble recording experiments demonstrated the critical nature of various components of the basal ganglia in temporal discrimination (e.g., Matell et al., 2003, Meck, 1996). Additionally, patients with Parkinson’s disease, who show deterioration of dopaminergic cell bodies within the substantia nigra pars compacta region of the basal ganglia, were found to display increased variability on a repetitive tapping task when they were withdrawn from levadopa treatment (O’Boyle, Freeman, & Cody, 1996). Taken together with the evidence for cerebellar dysfunction in SZ, the current literature suggests that patients with SZ have fundamental deficits within the neural structures that are essential for the execution of internal temporal processing.

In addition to the neural circuitry that has been implicated in timing processes, the dopaminergic neurotransmitter system appears to play an important modulatory role in the integration of temporal information across functional brain regions (Dolan et al., 1999, Meck and Benson, 2002). In a variety of time estimation tasks, dopamine (DA) agonists have been linked to over estimations of temporal intervals within the range of several seconds (Meck, 1983, Meck, 1996). In contrast, DA antagonists were found to produce under estimations of temporal durations, where decreased time estimation was proportional to the affinity of the DA antagonists for the D₂ sub-receptor (Meck, 1986, Rammsayer, 1997). Although these results may be reflective of a modulatory affect of DA on neural structures that are thought to be responsible for the encoding of time intervals (e.g., the basal ganglia), further evidence suggests that DA is also associated with attentional (Buhusi and Meck, 2002, Stanford and Santi, 1998, Tracy et al., 1998) and mnemonic (Rammsayer, 1997, Tracy et al., 1998) functions of the prefrontal and parietal cortex that are required to sustain temporal processing of durations beyond the range of milliseconds. Accordingly, DA abnormalities in cortical and subcortical areas have been linked to the pathophysiology of SZ (Hietala and Syvalahti, 1996, Scarr et al., 2001, Wolf et al., 1993), with attentional deficits receiving much support as one of the primary cognitive disturbances of the illness (Kimble et al., 2000, Nestor and O’Donnell, 1998). Thus, based on the evidence for the role of DA in temporal integration, in conjunction with the findings of cerebellar and basal ganglian dysfunction in SZ, it seems reasonable to suggest that at least some of the symptomotology associated with SZ may be a manifestation of abnormalities within the neural circuitry of internal timing mechanisms.

Eyeblink conditioning methodology is well suited for studying timing deficits in SZ. First, the neural circuitry associated with different forms of conditioning is distinct, well characterized, and includes structures and pathways implicated in SZ. All forms of EBC are dependent on cerebellar function and some forms are also dependent on the function of the hippocampus (Daum et al., 1991, Ross et al., 1984, Steinmetz, 2000). The importance of the cerebellum in the acquisition of the nictitating membrane/eyeblink response in rabbits has been clearly demonstrated (McCormick et al., 1981, McCormick and Thompson, 1984a, McCormick and Thompson, 1984b, Thompson, 1986, Yeo et al., 1985a, Yeo et al., 1985b). Although the neural substrates of eyeblink classical conditioning in humans are not as well characterized, recent evidence supports the critical role of the cerebellum in human eyeblink conditioning (Daum et al., 1993, Topka et al., 1993, Woodruff-Pak et al., 1996). For instance, a PET study showed activation in the inferior cerebellar cortex and deep cerebellar nuclei in humans during eyeblink classical conditioning (Logan & Grafton, 1995). Furthermore, cerebellar morphology and volume in both animals and humans is related to the magnitude of eyeblink conditioning (e.g., Woodruff-Pak, Goldenberg, Downey-Lamb, Boyko, & Lemieux, 2000).

Second, EBC methodology offers high temporal resolution and yields functional measures of associative learning (e.g., acquisition and extinction rates and response latencies). Both the cerebellar nuclei as well as the cerebellar cortex are required for accurate performance of the conditioned response. The cerebellar nuclei are critical to forming the necessary stimulus associations and compelling evidence indicates that anterior regions of the cerebellar cortex function to precisely time the learned response (Perrett, Ruiz, & Mauk, 1993). The connection between eyeblink conditioning and the timing functions of the cerebellum has been reported by Woodruff-Pak et al. (1996), who showed a negative correlation between conditioned responses (CRs) and a measure of clock variability in the timed-interval tapping task (Ivry et al., 1988) in healthy subjects. In addition, the variability in this interval timing task significantly predicted conditioning (Woodruff-Pak & Jaeger, 1998).

Finally, neurotransmitters that are implicated in SZ underlie EBC (DA—Sears & Steinmetz, 1990; glutamate—Fiala et al., 1996, Shors et al., 1995, Thompson et al., 1992). Specifically, DA has been shown to affect the speed of an internal timing mechanism. A recent study on the effect of haloperidol on the control of the internal clock found that haloperidol likely caused a deceleration of the internal clock, and a significant decrease in the attention given to gaps inserted into temporal signals that would normally cause subjects to stop timing (Buhusi & Meck, 2002).

The few extant studies of eyeblink classical conditioning in SZ have yielded equivocal findings and pose interpretive problems because of a lack of uniformity across studies and, in some cases, methodological inadequacies. Some of the earliest studies of EBC in SZ reported a reduction of CRs or no change compared to healthy, non-patient comparison participants (King and Landis, 1943, O’Connor and Rawnsley, 1959; see Lehmann & Ban, 1971), whereas others reported increased CRs in SZ (e.g., Spence & Taylor, 1954). In a study of single cue delay EBC, Spain (1966) reported that patients with SZ generated more conditioned responses than non-patient comparison participants; however, conditioning was associated with higher skin potentials, an index of arousal. When the SZ and comparison participants were matched on mean level of skin potential, there was no group difference in the number of conditioned responses. In addition, the conditioning abnormalities in SZ patients depended on the modality of the CS, with SZ patients showing facilitated conditioning to a visual conditioned stimulus (CS) but retarded conditioning to an auditory CS compared to the comparison participants. However, the fact that nearly one-third (17 of 54) of the SZ patients tested in the experiment were excluded from the analyses because of high rates of non-associative or spontaneous blinks further complicates the interpretation of these results.

More recently, Sears, Andreasen, and O’Leary (2000) reported that unmedicated patients with SZ (N = 15) exhibited accelerated acquisition of a conditioned eyeblink response in a single-cue auditory delay task. The findings were interpreted as evidence for enhanced excitability of associated cerebellar circuitry and consistent with a previous PET study showing greater cerebellar activation at rest in SZ compared to non-patient participants (Andreasen et al., 1997). Sears and colleagues speculated that this abnormally enhanced cerebellar plasticity predisposes individuals with SZ to form inappropriate associative connections, perhaps resulting in the characteristic behavioral disorganization and formal thought disorder seen in SZ. However, group differences in the rate of non-associative or spontaneous blinks were not examined as a possible explanation for the observed facilitated EBC in SZ. This seems especially important given the high rates of non-associative or spontaneous blinks observed in SZ by Spain (1966).

In the most recent study of single-cue auditory delay EBC, Marenco, Weinberger, and Scheurs (2003) reported that medicated SZ patients (N = 10) and non-patient comparison participants (N = 9) did not differ in the percentage of CRs, though the patients showed longer CR onset and peak latencies. In a second sample of patients (N = 10) and comparison participants (N = 10), Marenco and colleagues reported that neither group of participants learned the CR during a trace EBC task, in part because levels of spontaneous blink rates exceeded the level of the CRs. Finally, in a conditional-discrimination paradigm, Hofer, Doby, Anderer, and Dantendorfer (2001) reported that medicated patients with SZ showed impaired learning of the CR overall as well as poor discrimination. The authors noted that the inability of the SZ patients to discriminate might be accounted for by the fact that the conditional-discrimination paradigm involves more complex stimuli presentation, and therefore utilizes different neural circuitry than the single-cue delay paradigm used in other studies.

These incongruent observations warrant further examination. The present study examined single-cue tone delay eyeblink conditioning in a sample of medicated schizophrenia patients. It was hypothesized that SZ patients would show deficits in the acquisition of the conditioned blink response as well as greater within-subject variability in the timing of the conditioned blink responses.