Short Term Sleep Deprivation, Language Comprehension and Auditory Temporal Resolution

Transcription

1 ***This is a self-archiving copy and does not fully replicate the published version*** Short Term Sleep Deprivation, Language Comprehension and Auditory Temporal Resolution Harvey Babkoff1, Leah Fostick2, Gil Zukerman2 and Elisheva Ben-Artzi 3 1Department of Psychology, Bar-Ilan University, Ramat-Gan, Israel and Department of Psychology, Ashkelon Academic College, Ashkelon, Israel 2Department of Communication Disorders, Ariel University Center, Samaria, Israel 3Kinneret College, Tzemach, Israel

2 ABSTRACT Over the past several years there has been a shift in the theoretical construct or models used by sleep researchers to explain the effects of sleep deprivation on performance. The focus has changed from an emphasis on arousal mechanisms to that of prefrontal cortical activation and its associated cognitive functions. According to the latter theoretical construct, cognitive functions that are considered most vulnerable to short term sleep loss are language skills in communication, divergent thinking, creativity and cognitive flexibility. Although the theoretical framework for the impact of sleep deprivation on performance may have changed over time, nevertheless there has not been any increased interest in investigating the effects of sleep deprivation on sensory or perceptual functions. In the present chapter we develop the arguments for studying the impact of sleep loss on perceptual systems that are strongly associated with a cognitive function, such as language. Language comprehension and production have been strongly associated with auditory temporal resolution. Consequently we believe that performance on tasks based on auditory temporal resolution should be of paramount interest to researchers of the impact of short term sleep deprivation on cognitive functions. We present a reanalysis of auditory temporal resolution data published earlier to show how sleep loss impacts the ability to discriminate dichotic temporal order. The meaning of deficit in auditory temporal resolution and its possible relation to impairment in performing language-based tasks is not only theoretically interesting but may also be of practical use, e.g., when testing the effects of pharmacological or other putative treatments to ameliorate the effects of sleep loss on perceptual or cognitive performance.

3 INTRODUCTION Studies of sleep deprivation, to date, have generally concentrated on the effects of sleep loss on psychological, cognitive and motor performance (Babkoff, et al 1991a,and b; Babkoff, et al, 2005; Dinges & Kribbs, 1991). Very few studies have examined the effects of sleep loss on more basic sensory and perceptual functions (Babkoff et al., 2005). There was apparently a very good reason for the lack of interest in studying the effects of sleep loss on basic sensory and perceptual functions. In most of the earlier studies of the effects of sleep loss, the main interest was in designing performance tasks that were prima facie concordant with the "real life" complaints of individuals required to perform effectively under conditions of sleep deprivation. In general, these tasks tested: 1) for changes in mood, consistent with complaints of changes in emotional responses; 2) either had face validity with the tasks usually performed in the workplace; or 3) tested some very specific cognitive or psychomotor capability that has been shown to be a basic and necessary component of successful task performance, e.g., working memory, choice reaction time. Individuals suffering from insomnia or other sleep disturbances usually complain of fatigue and difficulty in performing their normal daily activities, rather than complaining of altered thresholds or other basic perceptual changes. Thus, superficially at least, there seems to be no meaningful negative impact of sleep loss on any of the basic sensory or perceptual functions that was worth investigating. Arousal Theory of Sleep Deprivation However, there is a second, perhaps more important, reason for the lack of interest in the impact of sleep loss on sensory and perceptual functions,

4 i.e., the theoretical framework within which researchers viewed and studied the impact of sleep deprivation on task performance. The prevailing thought regarding the brain mechanisms affected by sleep deprivation during the latter part of the twentieth century centered primarily on the arousal system. Up until the 1990s, many researchers designed their studies and interpreted their results in terms of the Arousal theory of sleep deprivation. They argued that a reduction in sleep length affects performance by causing decreased arousal and slower cognitive processing (Babkoff, et al, 1988, 1991a; Harrison & Horne, 2000a; Kjellberg, 1977; Wilkinson, 1965; Wilkinson, 1992; Williams, et el, 1959). According to this view, sleep deprivation serves to reduce the nonspecific arousal level, but has no specific effect on any given sensory, perceptual, or cognitive system (Wilkinson, 1992). The most extreme version of this hypothesis (Lapse hypothesis) proposed that the only real deficit due to sleep loss was an increase in errors of omission or lapses in performance (Williams et al., 1959). The Lapse hypothesis considered the lapse or an error of omission (defined behaviorally as a period of not responding that is longer than several standard deviations beyond the expected or mean response time) as the major dependent variable in studies of sleep loss and posited the presence of microsleeps (defined as an episode of sleep lasting from 0.5 to 10 seconds) during long hours of sleep deprivation as the major factor in performance decline during long term sleep deprivation. In its radical form, the Lapse hypothesis implies that sleep loss itself does not impact negatively on sensory, perceptual or cognitive functions. The only deficit, which is systematically found, lapses, is due to the subjects really having fallen asleep, physiologically, for short periods of time (i.e., subjects experience microsleeps). According to this view, sleep loss affects all cortical regions in a fairly similar manner.

5 Wilkinson (1992) even argued that sleep deprivation reduces the nonspecific arousal level, but has no specific effect on any sensory, perceptual or cognitive function differently from any other. Even in its milder form, Arousal theory posited a fairly uniform general slowing of responses in all neurobehavioral systems, rather than deficit in any specific system. Dinges and Kribbs (1991) noted that the majority of studies of sleep-deprivation and performance reported global changes in performance ability, changes that may have been mediated as much by microsleeps and lapses in attention as by more subtle or profound alterations in information processing. Moreover, in the early 1990's Koslowsky and Babkoff (1992) reviewed the results of 27 studies (from ) with a total of 429 subjects that met the criteria for a quantitative analysis, and applied meta-analytic techniques to test the lapse hypothesis. They reported that speed rather than accuracy of performance, and work-paced rather than self-paced tasks were most affected by sleep loss. The authors concluded that the findings were consistent with the Lapse hypothesis. In summary, according to the Arousal hypothesis, there were no intrinsic differences in performance decrement among different tasks and certainly not on any given sensory or perceptual system, since the major source of variance caused by sleep deprivation was basically the same for performance on all tasks, i.e., reduced arousal. The nature of the task itself and the sensory or perceptual systems involved were apparently of lesser importance. Although the major theoretical framework for studying and understanding the effects of sleep loss on task performance has shifted radically over the past decade, nevertheless the prevailing view seems to continue to view the study of the effects of sleep loss on sensory and perceptual functions as theoretically unimportant and practically

6 inapplicable to the workplace. In fact the emphasis of the more recent view in positing that sleep loss has specific as well as general effects on performance, has been on the higher cognitive level functions, rather than on the more basis sensory or perceptual systems. The major alternative hypothesis to that of Arousal, was proposed by Harrison and Horne (1998, 2000a, 2000b, 2000c) and has been termed the Sleep-Based Neuropsychological Perspective. Sleep Based Neuropsychological Perspective The 'sleep-based neuropsychological perspective' differs from the Arousal theory in that; 1) it does assume that sleep loss might affect different cognitive functions differently; and 2) it posits that sleep loss may affect different parts of the brain differently (Harrison & Horne, 1998, 2000a, 2000b). Harrison and Horne (2000a, 2000b) have argued that sleep reduction primarily impacts the functions associated with prefrontal cortex (PFC) activity. The prefrontal cortex is among the first brain regions to be affected by sleep deprivation (e.g., Drummond et al., 2000, 2001; Muzur, et al, 2002; Thomas et al., 2000, 2003). Thomas et al. (2000, 2003) reported progressive decreases in relative cerebral metabolic rate for glucose (CMRglu) in prefrontal and thalamic areas over the course of 72 hours of sleep deprivation. These changes were positively correlated with impairments in cognitive performance and saccadic velocity. Relative thalamic activity was also correlated with alterations in alertness. Altena et al. (2008) reported that insomnia interferes with activation of prefrontal cortical systems during daytime task performance. Some other recent work even indicates that prefrontal activation may not be fully restored by recovery sleep. Wu et al. (2006) studied sleep deprivation and recovery sleep in normal subjects and reported that sleep deprivation resulted in a significant decrease in the relative metabolism of the frontal cortex, thalamus, and striatum. They reported that recovery

7 sleep had only a partial restorative effect on frontal lobe function and suggest that sleep may be especially important for maintenance of frontal lobe activity. According to the Sleep Based Neuropsychological hypothesis, even with relatively mild to moderate sleep loss (24 hrs), tasks that are highly dependent on intact prefrontal cortical activity will show deficit. These researchers predict that deficit due to sleep deprivation mainly affects complex tasks that measure higher cortical related skills such as: language skills in communication, divergent thinking, creativity and cognitive flexibility. Cognitive and motor tasks demanding only basic skills, although requiring attention and arousal, should be relatively less affected after mild to moderate sleep loss. Does the shift in theoretical approach to a concentration on cognitive functions associated with intact prefrontal activation mean that studies of the effects of sleep loss on sensory and perceptual functions are of no real theoretical interest? In the present chapter, we argue that, even within the framework of the newer theoretical approach, there can be merit to studies of sensory and perceptual functions during sleep deprivation if the functions studied meet certain criteria: 1) the sensory or perceptual system studied has been shown to be strongly related to a cognitive function that is impacted by sleep deprivation; 2) that cognitive function is, itself, of theoretical and/or practical interest because of its necessity for optimal cognitive performance. In the following section, we present the argument that auditory temporal resolution meets the above enumerated criteria because of its very strong association with language and language-based skills, and is, therefore, a prime candidate for the study of the impact of sleep deprivation.

8 LANGUAGE and SLEEP DEPRIVATION The arguments and data analyses presented in the following section are formulated in a four-step logical structure. We: 1) review evidence that even short term sleep deprivation impacts negatively on language and language-based tasks; 2) review the evidence that relates auditory temporal resolution to intact language comprehension and/or production; 3) argue that consequently a psychophysical task that measures auditory temporal resolution, e.g., dichotic temporal order judgment (TOJ), should be negatively impacted by sleep deprivation; and finally 4) present evidence to support the argument. Language-based and language-dependent skills are the basic tools of our civilization and almost ubiquitous for successful work performance. The maintenance of a modern industrialized society depends upon the ability to communicate clearly and comprehensively. Today's work place is to a large extent, circadian-independent, operating around the clock. Furthermore, many of the indispensable services necessary for the maintenance of health and security must be available 24 hours a day, seven days a week. Very often this scheduling involves either around-the-clock shift work or extended work time at the expense of normal nocturnal sleep. Because of the major importance of intact language performance in an information-based modern industrialized society and given the requirement of continuous high level operations around the clock even under the most stressful conditions, it becomes important to study the effects of sleep deprivation on language comprehension and the neuropsychological and neurophysiological mechanisms involved in maintaining intact language skills. Indeed, earlier studies had reported that short term sleep deprivation may impair normal language generation and articulation (see e.g., Harrison & Horne,

9 1998). Sleep deprivation had also been reported to increase the number of inaccuracies in sending and receiving messages (Neville, Bison, French, & Boll, 1994; Schein, 1957).These findings may be taken to imply the expectation of difficulties in language comprehension as well as in language generation even after mild to moderate sleep loss. The inclusion of language-based tasks in studies of sleep deprivation is, by no means, a new phenomenon. Language-based tasks have been a part of many of the protocols of long term sleep deprivation studies for many years. For example, the effects of long term sleep deprivation on Braddely's grammatical transformation (logical reasoning) and working memory tasks were included in the test batteries of studies by a number of authors over 25 years ago (see, e.g., Angus & Heslegrave, 1985; Babkoff, et al, 1988; 1991a). More recently, however, as a result of the shift in the theoretical approach that emphasizes the impact of sleep loss on executive control processes and other cognitive skills that are highly related to prefrontal and frontal cortex, as noted above, the interest in the impact of even short term sleep loss or of reduced sleep on language and language-based tasks has increased, and become a topic of interest by itself. In a recent study, Pilcher et al. (2007) reported a complex impact of sleep loss on languagebased tasks. Specifically, Pilcher used a full and diverse series of language tests and language-based tasks of a wide range of complexity to study the effects of a combined 28-hour sleep deprivation and sustained operations schedule on non-native English speaking subjects. These tasks included, among others: sentence completion, word analogies, antonyms, reading comprehension and logical reasoning (taken from the Law School Admission Test). In addition, the authors tested their subjects on verbally presented stories from four non fiction audio books to which subjects listened and were required to identify a "keyword" and then after 25

10 minutes of listening summarize the information they had heard. Verbal working memory and the PVT (a 10-minute vigilance and reaction time) together with several additional tasks were also included as probes in the test battery. The authors reported that some, but not all, of the languagebased tasks were negatively impacted. Language tasks that required sustained attention or higher level processing, such as reading comprehension showed significant deficit whereas other tasks that were dependent on more basic language processing, e.g., antonym identification were not negatively impacted by the experimental conditions. Among the tasks significantly impacted by the sustained operations-sleep deprivation protocol were logical reasoning and a verbal working memory task. However, a number of the relatively short duration probe tasks, e.g., the verbal working memory, followed by the PVT tasks were good predictors of performance on eight of the ten language tasks used in the study. The authors concluded " that sustained work conditions and sleep deprivation negatively affect some types of language performance " Although there is fairly clear evidence from the Pilcher et al. (2007) study of the negative impact of short term sleep loss on language comprehension, there may, nevertheless be somewhat of a problem with interpreting their results, since their subjects were not only sleep deprived but also tested in a sustained operations protocol, with a relatively high work-to-rest ratio. Beginning with Test Session I at 1830 until the end of Test Session IV at 1200 the next day, their subjects were continuously occupied in the testing program except for three one-half hours of rest, for a total work time of 17.5 hours versus a total rest time of 1.5 hours. Thus the sustained operations component of the protocol (i.e., the very high work-to-rest ratio encompassing the testing period of over 10:1) may

11 be at least partially responsible for the deficits in performance rather than just the sleep deprivation of 28 hours (see e.g., Angus and Hestlegrave, Other studies, however, support the claim that short term sleep deprivation negatively impacts language-based tasks even when the work-to-rest ratio is low (Harrison & Horne, 1998; Linde & Berstrom, 1992; Monk & Carrier, 1997). For example, Monk and Carrier (1997) tested the performance of eighteen young adults during 36 hours of constant wakeful bed rest on a logical reasoning task. The authors reported that even after microsleeps, lapses in attention, and general slowing of motor responses were eliminated from the data, there was a significant slowing of cognitive processing. It took subjects longer to respond to items phrased in the negative voice than to items phrased in the positive voice. The extra cognitive processing took longer at night and on the day following sleep loss than it did during the day before the sleep loss. The authors interpret these results to mean that human mental processing slows down during the night under conditions of sleep loss even when lapses have been removed from the data. Harrison and Horne (1998) also tested subjects who were sleep deprived for 34 hours in a single session and reported impairment on the word fluency task and the Haylings test. In the latter test, the authors required their subjects to complete a sentence with a single word which was either (a) congruous, or (b) totally incongruous with the overall meaning of the sentence. Harrison and Horne (1998) suggested that the novelty component in certain language tasks may make those tasks more likely to be negatively affected by sleep loss. Taken together, it seems that overall there is sufficient evidence that even short term sleep deprivation can negatively impact language and language-based tasks, although it may not yet be clear what makes certain language tasks more sensitive than others to sleep loss, nor are the possible underlying mechanisms known. Certainly,

12 the mechanisms by which sleep loss affects language-based performance are, as yet, unclear. Auditory Temporal Resolution as a Marker of Intact Language Performance Over the past several decades there have been a large number of studies that have reported a strong relationship between auditory temporal resolution and intact speech comprehension in a variety of different populations (Farmer & Klein, 1995; Kiss, et al, 2008; Poeppel, 2003). These studies have compared individuals with normal language abilities to others with language difficulties that range from developmental difficulties to pathological neurological conditions to the difficulties reported by normally aging older individuals in comprehending speech. Poppel (1997) has argued that intact auditory temporal resolution is crucial to understand speech. Normal speech is produced at approximately 150 to 250 words per minute, i.e., 4 to 7 syllables per second (Goldman-Eisler, 1968; Huggins, 1964). Accordingly, syllable duration is 67 to 177 msec (Klatt, 1976), vowel duration is 70 to 130 msec (Klatt, 1976), and consonant duration is 5 to 40 msec (Kent & Reed, 1975). Rapid spectral changes such as formant transitions with information on place of articulation may occur within a temporal range of msec (Poeppel (2003) and syllabicity and prosodic phenomena are reported to occur within a temporal range of msec (Rosen, 1992). Furthermore,the temporal separations necessary to accurately identify two auditory stimuli (gap detection) and to correctly identify the temporal order of two stimuli in the spectral and dichotic TOJ tasks range between msec (Fostick, 2006; He, et al, 1999; Phillips, et al, 2000; Schneider & Hamstra, 1999; Snell, 1997; Snell & Frisina, 2000; Snell, et al, 2002; Strouse, et al, 1998).

13 Auditory Temporal Resolution and Language Development in Children Many developmental studies of children with language and reading difficulties as compared with their normal reading cohorts have included tests of auditory temporal resolution. The results, however, appear to be inconclusive: Some researchers believe that intact auditory temporal resolution is necessary for the development of normal reading ability (e.g., Tallal, et al, 1997). For example, Visto, et al (1996) compared children with specific language impairment (SLI) and children with normal language learning in their ability to process binaural temporal information. They reported that children with specific language impairment appear to be impaired in their ability to use non linguistic binaural acoustic information in a dynamic ongoing fashion. Visto et al. (1996) suggested that requirements for processing such nonlinguistic acoustic information in a "dynamic and ongoing" fashion may be similar to those involved in the ongoing processing of rapid changes in the temporal and spectral components of speech. Other researchers have concluded that intact auditory temporal resolution does not necessarily predict the development of good reading ability in children. McAnally, et al, (2004) compared children who were good readers and those that were delayed readers on the discrimination of differences in the temporal properties of sound sequences and concluded that there were no significant differences between the two groups. However, although researchers may disagree concerning the issue of auditory temporal resolution as a factor of causality in language disorders in children, all

14 researchers seem to agree that intact temporal resolution is an important marker of intact language function (Marshall, et al, 2001). Complaints of Language Difficulties and Auditory Temporal Resolution in the Healthy Normal Elderly Normal healthy aging often includes vulnerability to difficulties in language comprehension and auditory temporal resolution. A common complaint among the elderly is a difficulty in understanding speech, especially attempting to do so under less favorable conditions, such as when speech is accompanied by noise or when speech is rapid. Several theories have been presented to explain this age-related difficulty in comprehending speech. Since aging of otherwise healthy individuals is accompanied by changes in a variety of peripheral and central physiological, behavioral and cognitive functions, some of the theories have suggested that these changes are associated with the reported difficulty in understanding speech. The hearing-sensitivity theory argues that presbycusis and other age-related changes in peripheral hearing are the main causes for the difficulties in speech comprehension (Dubno, et al, 1984, 1997; Halling & Humes, 2000; Humes et al., 1994; Humes, 1996; Schneider, et al, 2002; Wingfield, et al, 2000). However, a number of reviews and studies have shown this explanation to be insufficient since correction for audiometric hearing loss does not fully explain the difficulties experienced by the elderly in speech comprehension (Drager & Reichle, 2001; Fitzgibbons & Gordon-Salant, 1996; Gordon-Salant & Fitzgibbons, 1993; He, et al, 1998; Snell, 1997; Tun, 1998). Over the past two decades, some researchers have focused on agerelated degradation in auditory temporal resolution as an explanation for speech comprehension difficulties observed in the elderly. The rationale underlying this hypothesis is that the appropriate use of speech cues relies on several types of auditory temporal resolution (Gordon-Salant, 2005;

15 Pichora-Fuller & Souza, 2003; Schneider & Pichora-Fuller, 2001; Schneider et al., 2002), which have been shown to decline with aging. For example, older adults perform poorer than younger adults in gap detection tasks and need longer silent intervals to identify the presence of a gap when the marker signal is 250 msec or shorter (Fink, et al, 2005; Fitzgibbons & Gordon-Salant, 2001; Grose, et al, 2006; Lister & Roberts, 2005; Lister & Tarver, 2004; Roberts & Lister, 2004; Schneider & Hamstra, 1999; Schneider, et al, 1998; Snell, 1997; Snell & Frisina, 2000; Snell et al., 2002; Strouse et al., 1998). Older subjects have difficulty in correctly identifying temporal order in a tonal sequence (Fitzgibbons & Gordon-Salant, 1998; Gordon-Salant & Fitzgibons, 1999). A number of studies have reported that older individuals require larger differences in duration between two tones in order to detect a difference in duration (Fitzgibbons & Gordon-Salant, 1994, 1996, 1998). Similar results, indicating poorer discrimination, were found when comparing older and younger adults on binaural temporal processing tasks such as locating a tone in the front-back plane (Abel & Hay, 1996), tone localization (Abel, et al, 2000), and click lateralization (Babkoff et al., 2002; Strouse et al., 1998). Auditory Temporal Resolution and Pathological Language Conditions in Adults Deficits in the processing of rapidly changing signals, as, for example, increased thresholds for the discrimination of the temporal order of auditory stimuli, are often correlated with phoneme- identification and phoneme-discrimination impairments in patients with left-hemispheric lesions to the brain and aphasia, as well as children and adults with dyslexia (e.g., Farmer & Klein, 1995; von Steinbuchel, et al, 1999; Swisher & Hirsh, 1972; Tallal & Piercy, 1973; Wittmann, et al, 2004). Based on these relationships, some authors have suggested that the

16 processing of rapid temporal changes in speech signals is required for correct phonemic processing (Efron, 1963; Fink et al., 2005; Poepel, 1997; Tallal, 1980; Tallal, 1984; Wittmann, 1999). Von Steinbuchel et al. (1999) and Fink et al. (2005) compared aphasia patients with normal controls on two measures of auditory temporal resolution, one of which was the dichotic temporal order judgment of two clicks (termed by the authors: "alternating monaural stimulation mode") and reported that patients with aphasia had significantly higher temporal order thresholds than the normal controls. In summary, there is increasing evidence of the strong relationship between deficits in language comprehension found in a variety of conditions, developmental, pathological and normal age-associated, and reduced auditory temporal resolution. A Study of Auditory Temporal Resolution and Short Term Sleep Deprivation Given the strong relationship between auditory temporal resolution and intact speech comprehension and the evidence that sleep deprivation impacts negatively on tasks requiring intact language comprehension as reviewed above, we attempted to address the question whether short term sleep deprivation impacts auditory temporal resolution negatively. We chose an experimental protocol that included a low ratio work-to-rest schedule environment (1.5 work: 1 rest) and tested auditory temporal resolution by measuring dichotic temporal order judgment (TOJ). In the paradigm used in our laboratory, dichotic temporal order judgments are measured by presenting a tone of a given frequency to one ear and the same frequency tone to the other ear separated by times ranging between approximately msec. Subjects are required to judge which of the two ears received the first stimulus. The reasons for choosing this paradigm to measure auditory temporal resolution are discussed below.

17 For the purpose of the arguments presented in this paper, we reanalyzed the data from our earlier study (Babkoff et al., 2005) combining the data of the first four sessions only, beginning 0830 and ending Data over that period of time represent hours of sleep deprivation. The Choice of Measure of Auditory Temporal Resolution in the Present Study As discussed above and expanded upon below, the hypothesis driving the present study was that mild to moderate sleep deprivation will affect dichotic TOJ negatively (Babkoff et al., 2005). The hypothesis is based on the following rationale. Intact auditory temporal resolution is strongly related to and may be a necessary component of intact language comprehension. Intact auditory temporal resolution requires normal activation of the prefrontal cortex. As discussed earlier, even short term sleep deprivation has been reported to impact negatively on functions requiring normal activation of prefrontal cortex and on some languagebased tasks. Dichotic TOJ is a valid measure of auditory temporal resolution. Therefore, we hypothesized that 24 hours of sleep loss should impact negatively on dichotic TOJ. But why choose dichotic temporal judgments (TOJ) to represent auditory temporal resolution when studying the effects of sleep loss since there are a number of other procedures that have been used by researchers over the past several decades to measure auditory temporal resolution? In fact, auditory temporal resolution has been measured by: 1) the ability to discriminate compressed speech (Versfield and Dreschler, 2002); 2) gap detection (Schneider and Hamstra, 1999); 3) the discrimination of stimulus duration (Fitzgibbons and Gordon-Salant, 1994); 4) localization and lateralization (Babkoff, French, Whitmore, & Sutherlin 2002), as well as temporal order judgment (Ben-Artzi, Fostick, & Babkoff, 2005:

18 Farmer & Klein, 1995). There are several reasons why we preferred the dichotic TOJ procedure over the aforementioned procedures especially for measuring the effects of short term sleep deprivation on auditory temporal resolution. First, the study was designed to use a protocol for which there is evidence that the procedure tests an auditory temporal resolution phenomenon in the tens to hundred msec range. Resolution in this temporal range is crucial for decoding and understanding language (see discussion above). Since dichotic temporal order judgment thresholds are most often reported in the tens to hundred msec range (see e.g., Ben-Artzi et al., 2005), such data would represent auditory temporal resolution in the same time range as that necessary for intact language comprehension and perhaps provide a prime facie relationship between resolution in this temporal range and intact language comprehension. Second, our study was designed to use a protocol for which there is evidence that the experimental procedure clearly tests central auditory temporal processing, rather than one that might be strongly influenced by peripheral auditory processing. For example, one of the more popular and often used protocols to measure temporal order judgment uses the following procedure. Two tones of different frequencies (one high and one low) are presented either to one ear or to both ears separated in time, and the subjects are required to report whether the order of presentation was high-low or low-high (Ben-Artzi et al., 2005; Hirsh, 1959; Hirsh & Sherrick, 1961; Reed, 1989; Tallal, 1980, 1984). The identification of the tones depends upon the difference in frequency between the two. However, two stimuli that differ in frequency, are presented to the same ear and are separated by very short inter-stimulus intervals often give rise to spectrally discriminable sequences when the order is reversed. The result is that discrimination by the subjects of the temporal order of the

19 two tones, high and low, may be based not only on the temporal resolution, but also on the spectral resolution of the auditory nervous system that begins with the interaction of the two stimuli at the peripheral level, i.e., at the basilar membrane of the same ears(s) (see Ben-Artzi et al, 2005 for a full discussion). Warren (1974, 1976, 1993) had noted the difficulty in using spectral temporal order to assess temporal resolution in a completely independent manner because of the confounding presence of the temporal envelope cue. In contrast, dichotic TOJ is one of the only procedures for which the anatomy of the auditory system guarantees that the two stimuli used (one to each ear) are processed together only in the central nervous system (brain stem) where encoded information from the two ears interact for the first time (see Pastore & Farrington, 1996, for review). Consequently, to study the effects of short term sleep deprivation on auditory temporal resolution, we used a variation of the temporal order judgment task, in which pairs of equal-frequency tones are presented dichotically with varying inter-stimulus intervals separating their arrival time at the two ears (left right or right left). In the dichotic form of the temporal order judgment (TOJ) paradigm, the location of the two stimuli (i.e. left or right) distinguishes them so that their order may be judged, although the two stimuli in each pair are identical in frequency. Some authors who have also used a similar temporal order judgment paradigm have termed the procedure: "alternating monaural stimulation mode" (Fink, et al, 2006). Third, the study was designed to use a procedure for which there is evidence of prefrontal cortex involvement. There is accumulating evidence of the involvement of areas in the prefrontal cortex (PFC) in the comprehension of rapidly changing speech and in the resolution of nonspeech auditory stimuli (Temple et al., 2000). Several researchers have argued that speech signals that are stable or that change only slowly in

20 time (over hundreds of milliseconds) appear to be processed bilaterally. However, speech signals that change rapidly (over tens of milliseconds) appear to be preferentially processed by left hemisphere mechanisms (Johnsrude, et al, 1997). Johnsrude et al. (1997) reported that when normal adults discriminated auditory stimuli with frequency glides of either long or short durations (30 ms versus 100 ms) there was significant cortical activation foci in the left orbito-frontal cortex, the left fusiform gyrus and the right cerebral hemisphere. Temple and Posner (1998) reported an increase in fmri responses in left inferior frontal cortex with increasing speech compression. Temple et al. (2000) also reported increased left dorso-lateral prefrontal activation in response to rapidly changing non-linguistic acoustic stimuli. They concluded that although left superior temporal gyrus seems to be preferentially involved in processing meaningful stimuli, the left prefrontal region is involved with rapidly changing non-verbal acoustic stimuli as well. Their findings and those of Belin et al. (1998) emphasize the role of the left prefrontal region in processing linguistic as well as non-linguistic rapidly changing acoustic stimuli. Joanisse and Gati (2003) found regions in the posterior portion of the left and right superior temporal gyrus and the left inferior frontal gyrus that show increased activation to tasks requiring rapid temporal discrimination, whether the stimuli were speech or non-speech auditory signals. They concluded that there is significant overlap in the brain regions involved in processing the rapid temporal characteristics of both speech and non-speech stimuli. A Summary of the Study: Procedure and Results See Babkoff et al (2005) for a full description of the study, including choice of subjects, procedure and method. In summary, seven men and thirteen women (mean age= 23.8 yrs), who reported no psychiatric or sleep disorders served as subjects. None of the subjects

21 reported consuming more than three cups of coffee or any other caffeinecontaining beverage or smoking more than one cigarette per day. The Study One week separated each of the two testing days. One test day followed normal nocturnal sleep, the other test day followed 24 hours of sleep deprivation. Subjects spent the night prior to testing at home. Subjects were told that during the night they were sleep deprived they could read, watch TV or videos, but must refrain from physical exercises, smoking or ingesting any foods or beverages that contained caffeine or alcohol. In addition, they were to remain at home. Subjects were also instructed that on the sleep deprivation night, they were to telephone the lab every 20 minutes beginning from 2000 until 0600 the next morning and to leave a short message verifying they were awake. Compliance with the stay-at-home instruction could be monitored by noting the phone number used to relay the message. Subjects were also told that if they failed to call even once during the sleep deprivation night, they would not be tested the following day. The 20 subjects who were tested complied fully with the instructions. Subjects were paid the equivalent of 950 NIS (approximately $210) for complete participation in the experiment. The average TOJ performance data re-analyzed and reported in this chapter are from the first four sessions of each of the testing days (Babkoff et al, 2005). The four sessions took place at approximately 0830, 1030, 1245 and 1530 on each of the two test days. In addition to the formal testing sessions, all subjects were familiarized with all of the tasks in a full training session that took place prior to the first testing day. All testing took place in a sound isolated testing room. A cognitive testing session took approximately 70 minutes, of which the TOJ testing procedure lasted approximately minutes. Because of the difference in the duration of the testing sessions, due to the addition of

22 electrophysiological recordings at three of the sessions, the inter-session intervals were not equal in length. The work-to-rest ratio over the four sessions was approximately 1.5:1 (1.5 hours of testing for every hour of rest). TOJ Testing Procedure and Stimuli. The dichotic TOJ testing procedure in the present study was identical to the procedure used in an earlier study (Ben Artzi et al, 2005), except that the duration of the stimuli in the present study was 10 msec. Each session consisted of three phases: 1) the preparation and familiarization phase; 2) the training phase; and 3) the formal testing phase. Subjects were initially presented with six examples of each one of the two 10 msec duration tones, one high frequency (1.5 khz), and one low frequency (1 khz) to identify. Then on the next 24 trials, the two tones were presented to one of the two ears (right or left) and the subject indicated whether the tone was presented to the right or left ear (while disregarding the frequency of the tone) by pressing a key on the computer keyboard. Visual feedback ( correct/incorrect ) was provided after each response. At the beginning of the training phase, the subjects were presented with the stimuli at one of the two ears in random order and required to identify the ear receiving the stimulus. No feedback was provided. Training continued until a criterion of 8 correct responses out of 10 trials were met (Binomial test, p<0.01). Any individual unable to meet this criterion was dismissed from further participation. The training then continued with the presentation of 16 dichotic tone pairs each consisting of 10 msec duration tones of the same frequency (e.g., 1 khz or 1.5 khz) and presented with an inter-stimulus interval (ISI) of 240 msec. Subjects were required to indicate the order in which the tones were presented (either right before left or left before right) by first pressing a key

23 indicating the first ear receiving the tone and then pressing a second key indicating the second ear receiving the tone. Feedback was provided for all of the 16 trials. The formal testing phase followed and consisted of 576 trials, on each of which a dichotic tone pair of the same frequency (either 1 khz or 1.5 khz) was presented with either the first member of the tone pair to the right ear followed by the second tone of the pair to the left ear or the reverse. On each trial the two tones of each pair were separated by one of the following inter-stimulus intervals (ISI): 5, 10, 15, 30, 60, 90,120, 240 or 400 msec. The ear receiving the first tone and the ISI were randomized by trial. Thus there were 64 trials at each ISI, i.e., 32 trials for each of the two frequency pairs (1kHz and 1.5 khz) on each session for each subject. Accuracy and RT were recorded on each trial. No feedback was provided during the formal testing phase. Since there was no significant difference between dichotic temporal order judgments to tone pairs of 1-kHz or 1.5- khz, the data were combined for all further analyses. Dichotic Temporal Order Judgments (TOJ Thresholds) TOJ was assessed during each of six diurnal sessions during a waking day following: 1) normal nocturnal sleep and 2) after 24 hours of sleep deprivation. Eighteen of the 20 subjects successfully completed the first four sessions (0830, 1030, 1245 and 1530) under both the non-sleep deprived and the sleep-deprived conditions. The data analyses therefore are based on the eighteen subjects and include only the first four sessions of each day (after normal sleep and after 24 hours of sleep deprivation. The Greenhouse-Geisser correction was applied to all tests of significance. A polynomial equation was fitted to each of the accuracy-isi curves for each subject under each of the eight conditions (four sessions when non sleep deprived and four sessions when sleep deprived) to assess

24 the dichotic TOJ threshold. Dichotic TOJ threshold was defined as the ISI at which the accuracy-isi best fitting polynomial curve crossed 75% correct. The threshold values were only assessed with interpolated, but not extrapolated values, so that no thresholds were either shorter than 5 msec or longer than 400 msec. All of the eighteen subjects who completed four full sessions under the non sleep deprivation and the sleep deprivation conditions, yielded psychophysical accuracy-isi curves after averaging, with levels greater than 75% correct at the longer ISI values so that a TOJ threshold could be evaluated as defined above. The dichotic TOJ thresholds of the first four sessions were averaged for each subject and the thresholds generated under the non sleep deprived condition were compared with those generated under the sleep deprived condition. Figure 1 is a histogram depiction of the distribution of dichotic TOJ thresholds for the non sleep deprived and the sleep deprived conditions averaged over the four sessions. The dichotic TOJ threshold was longer under the sleep deprivation condition (proportion greater than 1) than under the non-sleep deprivation condition for fourteen of the eighteen subjects (Binomial distribution, p<0.03) Fig. 1 about here The variance of the dichotic TOJ threshold distribution under sleep deprivation, however, was significantly larger (see Fig. 1) than that of the non-sleep deprivation condition (F (1, 17) = 3.93p < 0.004). Two types of data analyses were performed. First, the non-sleep deprived dichotic TOJ thresholds were compared with the sleep deprived dichotic TOJ thresholds by the non-parametric Wilcoxon Signed Ranks test for related data. Dichotic TOJ thresholds under the sleep-deprived condition were

25 significantly longer than under the non-sleep deprived condition (Z= ; p<0.01). Second, the mean dichotic TOJ thresholds from the four diurnal sessions on which a threshold could be assessed were log transformed and were compared by a one-way repeated ANOVA (NSD/SD). The results confirmed the significant difference between the assessed thresholds under sleep deprivation and non sleep deprivation (log threshold NSD= , sd=.29157; log threshold SD= sd=.3297; F(1,17)=9.193; p< 0.008; eta2=.35). The geometric means of the non sleep derived and sleep deprived dichotic TOJ threshold conditions were 63.4 msec and 84.2 msec respectively. If, as proposed and developed above, dichotic TOJ is a measure of the resolving power of the auditory temporal domain, then these data indicate that hours of sleep deprivation decrease that resolving power by approximately 32.8%. Auditory Temporal Resolution and Hours of Sleep Deprivation The meaning of 32.8% reduction in auditory temporal resolution and the possible relationship to potential difficulties in language comprehension may be understood better by comparing a reduction in resolution of this magnitude with similar reductions in temporal resolution, also measured by dichotic TOJ, reported in a variety of populations whose symptoms include a variety of language difficulties. The purpose is for comparison only, with no pretension of implying any similarity in mechanism. The data presented in Table 1 allow one to compare the extent of the reduction in auditory resolution measured by dichotic TOJ in pathological conditions to the extent of reduction experienced by an individual who was sleep deprived for hours. For example, Szymaszek,et al (2006) compared a group of young subjects with a group of older subjects on the dichotic (their alternating monaural condition)

26 TOJ thresholds of click stimuli. They reported average dichotic TOJ thresholds of 66 msec for the young subjects and 88 msec thresholds for the elderly, a reduction in resolution of 33%. In a recent dissertation, Fostick (2006) also compared a group of young subjects with a group of older subjects on the dichotic TOJ thresholds of 10 msec duration tones and found an average of msec thresholds for the young and msec thresholds for the elderly subjects, an estimated reduction in auditory temporal resolution of 30.3%. Kinsbourne et al. (1991) compared TOJ in normal and dyslexic adult readers, using 1 msec click stimuli presented dichotically. They used an ascending method of limits to determine a 90% threshold, which they reported to be 46.8 msec for the normal readers and 67.4 msec for the dyslexic readers, a 44% difference in temporal resolution for the adult dyslexics relative to the normal readers. Ben-Artzi et al. (2005) compared the dichotic TOJ thresholds of normal adult readers with those of adult dyslexics using 15 msec, 1000 Hz tones as stimuli. The average dichotic TOJ threshold of the normal adult readers was msec, while that of the adult dyslexics was msec, a reduction of ~65% in auditory temporal resolution in adult dyslexics as compared with the normal readers. In a recent dissertation, Bar-El (2009) compared the dichotic TOJ of adult normal and dyslexic readers and found an average threshold of msec for the normal readers and msec for the dyslexic readers, a reduction of 65 % in auditory temporal resolution. Von Steinbuchel et al. (1999) compared dichotic TOJ in normal adults with patients suffering from unilateral focal brain lesions, localized in anterior or posterior regions of the left hemisphere (LH) (with symptoms of non-fluent or fluent aphasia, respectively), or in predominantly sub-cortical regions of the LH (without aphasic symptoms) and in the anterior or posterior regions of the RH. Dichotic

27 TOJ threshold was found to be 57.7 msec for the normal population and msec for the patients with fluent aphasia, a reduction of ~ 104% in auditory temporal resolution. In summary, it appears that the estimated average reduction in auditory temporal resolution as measured by dichotic temporal judgment thresholds due to hours of sleep deprivation is numerically quite similar to the reported reduction in auditory temporal resolution of the elderly as compared to young subjects, around 30%. If we rank the amount of deficit in dichotic TOJ in the various populations and conditions reviewed, it appears that deficit in dichotic TOJ in the elderly and light to moderately sleep deprived young individuals ranks lowest, around 30%, followed by deficit in the dyslexic that ranges between 44-65%, followed by individuals with brain lesions and aphasia, who show deficits of 100% and over, i.e., the need for more than twice the dichotic inter-stimulus interval before the correct order of stimuli can be judged. Auditory Temporal Resolution and Language Comprehension: A Suggestion for Future Research Without assuming any identity or even relation of mechanisms, but just arguing by analogy, the similarity between the 30% deficit in auditory temporal resolution of the elderly and of younger individuals who experienced mild to moderate sleep loss may suggest a further direction of research. The major difficulty of the elderly in understanding speech seems to occur when speech is somewhat degraded or the conditions are less than favorable, e.g., when speech is accompanied by noise or when speech is rapid. As discussed above, the rationale underlying the hypothesis that relates auditory temporal resolution to language comprehension deficit in the elderly is based on the theory that the appropriate use of speech cues relies on several types of auditory