Journal of Neuroscience Methods 259 (2016) 57–71

Contents lists available at ScienceDirect

Journal of Neuroscience Methods journal homepage: www.elsevier.com/locate/jneumeth

Basic neuroscience

A rodent model of the human psychomotor vigilance test: Performance comparisons Catherine M. Davis a,∗ , Peter G. Roma a,b , Robert D. Hienz a,b a Division of Behavioral Biology, Department of Psychiatry and Behavioral Sciences, The Johns Hopkins University School of Medicine, Bayview Medical Center, 5510 Nathan Shock Drive, Suite 3000, Baltimore, MD 21224, USA b Institutes for Behavior Resources, Baltimore, MD, USA

h i g h l i g h t s • • • • •

Analogous to the human PVT, the rPVT is an effective task for preclinical studies. Describes the design and empirical validation of a novel PVT for use with rats. Results demonstrate effectiveness of the rodent PVT (rPVT) for assessing attention. Amphetamine increases while zolpidem decreases rPVT performances in rats. The rPVT is sensitive to radiation-induced deficits in attention.

a r t i c l e

i n f o

Article history: Received 21 August 2015 Received in revised form 17 November 2015 Accepted 19 November 2015 Available online 27 November 2015 Keywords: PVT Attention Vigilance Radiation Circadian disruption Rat Time on task Response-stimulus interval effect Vigilance decrement Sleep rPVT

a b s t r a c t Background: The human Psychomotor Vigilance Test (PVT) is commonly utilized as an objective risk assessment tool to quantify fatigue and sustained attention in laboratory, clinical, and operational settings. New method: Recent studies have employed a rodent version of the PVT (rPVT) to measure various aspects of attention (lapses in attention, reaction times) under varying experimental conditions. Results: Data are presented here to evaluate the short- and long-term utility of the rPVT adapted for laboratory rats designed to track the same types of performance variables as the human PVT—i.e., motor speed, inhibitory control (“impulsivity”), and attention/inattention. Results indicate that the rPVT is readily learned by rats and requires less than two weeks of training to acquire the basic procedure. Additional data are also presented on the effects of radiation exposure on these performance measures that indicate the utility of the procedure for assessing changes in neurobehavioral function in rodents across their lifespans. Comparison with existing method(s): Once stable performances are obtained, rats evidence a high degree of similarity to human performance measures, and include similarities in terms of lapses and reaction times, in addition to percent correct and premature responding. Similar to humans, rats display both a vigilance decrement across time on task and a response-stimulus interval effect. Conclusions: The rPVT is a useful tool in the investigation of the effects of a wide range of variables on vigilance performance that compares favorably to the human PVT and for developing potential prophylactics, countermeasures, and treatments for neurobehavioral dysfunctions. © 2015 Elsevier B.V. All rights reserved.

1. Introduction The human psychomotor vigilance test (PVT) is a widely validated and broadly applied assay of vigilant attention and basic

∗ Corresponding author. Tel.: +1 410 550 2775; fax: +1 410 550 2780. E-mail addresses: [email protected] (C.M. Davis), [email protected] (P.G. Roma), [email protected] (R.D. Hienz). http://dx.doi.org/10.1016/j.jneumeth.2015.11.014 0165-0270/© 2015 Elsevier B.V. All rights reserved.

neurocognitive function. It is partly rooted in the simple reaction time (SRT) procedure that has a long history in human psychology, starting back in the German laboratory of Wilhelm Wundt in late 19th century, and continuing on into the early 20th century at the Columbia laboratory of Cattell (1947). The human PVT as originally developed by Dinges et al. (1987), Dinges and Powell (1985), Kribbs et al. (1993), however, differs substantially from typical SRT procedures in terms of its procedural emphasis on assessing reaction time stability as well as general performance stability

58

C.M. Davis et al. / Journal of Neuroscience Methods 259 (2016) 57–71

(e.g., errors of commission and omission) across time within individual sessions. The modern PVT has been greatly refined over the years by Basner and Dinges (2011), Basner et al. (2011), Dinges et al. (1997), Drummond et al. (2005), Jewett et al. (1999), Lim and Dinges (2008a), Van Dongen and Dinges (2005), Van Dongen et al. (2001) as a human cognitive neurobehavioral assay for tracking temporally dynamic changes in sustained attention, and has been shown to be sensitive to sleep deprivation, fatigue, drug use, and age (Blatter et al., 2006; Lim and Dinges, 2008b). The PVT is a deceptively simple procedure that requires a subject to touch a screen when a stimulus (typically an LED counter) appears after 2–10 s, with the counter being incremented in milliseconds and stopped when the subject touches the screen, thus displaying to the subject his/her reaction time (RT) to the stimulus onset. The PVT reliably tracks fatigue-related decrements in vigilant attention as shown by a slowing in reaction time, an increase in “lapses” (errors of omission, typically defined as RTs > 500 ms), and an increase in errors of commission (“false starts”, or premature responses prior to the stimulus onset); however, other additional measures, including the fastest 10% of RTs (Q-10), the slowest 10% of RTs (Q-90), median RTs (Q-50), and mean RTs, can be acquired with the PVT and have been used to investigate various parameters such as gender and age differences (for a review, see Basner and Dinges, 2011). Further, the PVT has been used in human risk assessment in a range of operational environments (e.g., the military, the aviation and railway industries, first responders) and also employed in extreme environments such as NASA’s Extreme Environment Missions Operations (NEEMO), the international Mars500 Project (Basner et al., 2013), and on the International Space Station (ISS) where it is referred to as the “Reaction Self-Test” and provides astronauts with individualized performance feedback. While SRT procedures have been used for decades in animal research to examine a variety of sensory and motor functions (see Moody, 1970), animal versions of the human PVT have only recently begun to appear. One of the earliest uses of a human SRT procedure adapted for animals was provided by Skinner (1946), who trained pigeons with a “ready” or alerting signal to indicate the subsequent occurrence of a “reaction time” stimulus. Short-latency responses to the reaction time stimulus were additionally differentially reinforced. With this procedure he was able to obtain latencies in the range of 200–300 ms (see Moody, 1970). Since that time, numerous versions of the SRT procedure have been used in a wide range of animal research which in general may be subdivided into “signaled” (containing an alerting or “ready” signal) and “unsignaled” (i.e., no alerting signal) RT procedures, with the latter types being in essence analogs of the human PVT. SRT’s in rats, for example, typically consist of training rats to respond on a manipulandum (e.g., pressing a response lever with a paw, poking a lighted key with the nose, breaking a photo beam with the head) when a cue light is randomly illuminated, and to refrain from responding in the absence of the cue light (Baunez et al., 2001; Brown and Robbins, 1991; Domenger and Schwarting, 2006; Eckart et al., 2012; Li et al., 2010; Mayfield et al., 1993; Muir et al., 1996; Phillips and Brown, 1999; Pirch, 1980; Smith et al., 2010; Ward et al., 1998). Despite the human PVT’s decades of demonstrated utility and popularity, a direct rodent counterpart was first reported in the literature by Christie and colleagues who developed a version for rats and demonstrated that it tracks the same types of performance variables as the human PVT – e.g., general motor function and speed, premature responding and lapses in attention – and that it also is sensitive to decreased vigilance following sleep deprivation (Christie et al., 2008a, 2008b; for more recent versions of the rPVT, see also Deurveilher et al., 2015; Loomis et al., 2015; Oonk et al., 2015). While these reports have demonstrated the utility of the rPVT procedure in assessing the effects of sleep deprivation on sustained attention, the rats in many of these studies emitted large

numbers of premature responses that frequently made up more than 40% of the total number of responses, which is quite unlike any typical human PVT performance. This difference may be due to the specific parameters employed in the human vs. the rodent PVT; for example, the Christie et al. version of the rPVT used a variable 3–7 s foreperiod, compared to a human PVT that typically uses a 2–10 s foreperiod (although there is a 3-min version of the human PVT that uses a 1–4 s foreperiod; see Basner et al., 2011). Such relatively short variable foreperiod values may have promoted the increased numbers of premature responses reported in many of these rodent rPVT studies. The version of the rPVT described in the current study improves upon the previously-published rPVT by training rats to a greater level of behavioral control by (1) the use of variable foreperiod values between 3 and 10 s that more closely mimic the values used in the human PVT (i.e., 2–10 s); (2) the use of a short response window following stimulus onset (referred to below as the limited hold; 1.5 s in the present study compared to 3.0 s in the previous studies); (3) the demonstration of predictable changes in performance metrics that parallel those seen in humans when examining the vigilance decrement (Lim et al., 2010); (4) demonstration of changes similar to the variable response-stimulus interval (RSI) effect seen in the human PVT (Tucker et al., 2009) within the 3–10 variable foreperiod; and (5) the dissociation of these latter two metrics as previously reported for the human PVT (Tucker et al., 2009). As an additional step in validating the rPVT as a rodent model for assessing neurobehavioral function, the present study provides normative animal performance data using the rPVT as well as further demonstrations of the sensitivity of the rPVT to the long-term effects of radiation exposure on the CNS, to the effects of acute drug injections, and to circadian disruptions. 2. Methods 2.1. Subjects and apparatus Over the last six years, approximately 500 rats have been trained on the rPVT procedure in the laboratory. For the present report, data are presented for 122 male Long-Evans rats exposed to an automated training program that gradually shaped each rat’s behavior until the final rPVT performance was established. Data are also reported for 5 previously trained female Long-Evans rats for general performance comparisons between males and females. All rats were acquired at approximately 12 weeks of age, and were housed individually under a 12:12 h light/dark cycle (lights on at 0600 h) with continuous access to water and with food freely available. Animals were allowed to free-feed until their weights approximated the 340 to 350 g range (235–250 g for females) at which body weights were maintained for the following behavioral studies (Ator, 1991). Under the rPVT procedure, rats earned food (45-mg Noyes Precision rat pellets) during the experimental sessions, and were supplemented with commercial laboratory rat chow after the sessions to maintain their weight. When sessions were not conducted, the rats were fed 10–20 g of the rat chow, which resulted in weight stability or weight gain on the day the rats were next weighed. Extra food was also provided on weekends, or when no experimental sessions occurred. All rats were run at the same time of day by use of identically constructed experimental chambers (Med Associates® ). The front wall of each chamber contained 1 backilluminated nose-poke response key on the left, an overhead house light, and a food cup on the center for delivery of food pellets. All chambers were enclosed in sound-attenuating chambers equipped with an exhaust fan. Experimental contingencies were controlled by MedPC behavioral control programs running on PCs; the programs recorded all data on a trial-by-trial basis to provide for a wide range of subsequent analyses.

C.M. Davis et al. / Journal of Neuroscience Methods 259 (2016) 57–71

59

Fig. 1. Diagram of the rPVT procedure. The basic response measures of the rPVT are indicated by the boxes and arrows with solid lines, and are described in the text under “Basic Measures”. The boxes and arrows with dotted lines represent additional ITI responding and lapse measures, and are described in the text under “Additional Computed Measures”.

2.2. Ethical statement Laboratory animal care was provided according to Public Health Service (PHS) Policy on the Humane Care and Use of Laboratory Animals, and the Institutional Animal Care and Use Committee of the Johns Hopkins University approved all procedures. Johns Hopkins also maintains accreditation of their program by the Association for the Assessment and Accreditation of Laboratory Animal Care (AAALAC). 2.3. Rat psychomotor vigilance test (rPVT) 2.3.1. Basic procedure Each rPVT session began with the onset of the house light (see Fig. 1). After a variable interval of 3–10 s, the light behind the nose-poke key was illuminated for 1.5 s. In SRT procedures, this variable interval is often referred to as the variable foreperiod (i.e., the variable waiting time before a stimulus is actually presented). Foreperiods were randomly generated without replacement from a list of 36 possible foreperiod values between 3 and 10 s (based on 200 ms increments of the values). Thus all foreperiod values were sampled every 36 trials. A correct response was defined as a response on the nose poke key within the interval from 150 to 1500 ms after the light onset, following which both the nose-poke key light and the house light were turned off and a 45-mg food pellet was delivered as reinforcement. Nose-pokes prior to or within 150 ms after light onset were defined as premature responses, were not reinforced, and resulted in an 8 s timeout (TO) from the experimental contingencies (signaled by extinguishing the house light). If the 1.5 s interval elapsed without a response, both the key light and house light were turned off and the trial was defined as a miss. A 1 s inter-trial interval (ITI, during which the house light was off) followed either a correct response or a miss. The variable foreperiod for the next trial began after either the 1 s ITI or the 8 s TO, whichever occurred during the prior trial. Daily sessions consisted of approximately 200 trials and lasted approximately 30 min. 2.3.2. Training procedure Each rat was first shaped by the method of successive approximations (Skinner, 1938) to touch a lighted nose poke key with its nose, which normally required one to three 30-min sessions. Once an animal completed a minimum of 25 responses on its own, it was placed on an rPVT shaping program designed to slowly move each animal’s performance toward the final performance of the rPVT procedure as described above. Initially, shaping trials consisted of the key light coming on 2 s after the 1 s ITI, and remaining on for 9 s, allowing ample opportunity (i.e., a 2-s to 9-s response window) for

a reinforced response to occur. Subsequent changes in the foreperiod window were dependent upon an animal’s performance such that, when an animal achieved 8 correct responses (i.e., responses within the 2 s to 9 s response window) within a 10-trial span, the 2 s foreperiod was increased by 0.1 s to 2.1 s. If an animal did not reach this 80% correct criterion, the foreperiod remained at 2 s for the next 10 trials. Thus the foreperiod was successively increased in 0.1-s increments throughout a session, dependent upon an animal responding at an 80% correct level at each increasing foreperiod value. At the end of each training session, the program displayed the last successful foreperiod value achieved, and the next day’s session was started at this last successful foreperiod value. The training program concurrently modified other rPVT parameters. The TO value was slowly increased by making it equal to the current foreperiod value until a TO of 8 s was reached, whereupon it was fixed at 8 s. Additionally, the correct response window was slowly narrowed by reducing the upper limit of 9 s by 0.1-s each time the foreperiod was increased until a response window upper limit of 1.5 s was reached, whereupon it was fixed at 1.5 s. Once a rat was responding at the 80% correct level at a foreperiod length of 9.5 s, the shaping program switched over to presenting randomly-generated foreperiods. Initially, the foreperiod was varied between 7 and 10 s. Once an animal achieved 80% correct responding for a minimum of 30 trials at these values, the foreperiod range was automatically changed to 5–10 s. Finally, once an animal achieved 80% correct responding for a minimum of 30 trials at these values, the final foreperiod range of 3–10 s was put in place. 2.4. Response measures Basner and Dinges (2011) previously outlined the most commonly reported human PVT outcome metrics and from most to least frequently reported, include (1) number of lapses, (2) mean RT, (3) mean 1/RT (referred to below as ‘speed’), (4) fastest 10% RT (referred to below as Q-10), (5) median RT (also referred to as Q-50), (6) slowest 10% RT (referred to as Q-90 below), (7) slowest 10% 1/RT, (8) number of false starts, (9) fastest 10% 1/RT, (10) lapse probability, and (11) other. For the rPVT described here, these measures are described, in comparison to their PVT counterpart, in Table 1. Similar to human RT studies, the rat RTs were transformed for the various speed measures; Whelan (2008) argues that “transforming RTs to speed (i.e., the reciprocal of latency) normalizes the distribution somewhat, reduces the effect of slow outliers, and therefore generally maintains good power”. The rat performance measures are described in two sections below: basic measures that are acquired during the behavioral session, and computed

60

C.M. Davis et al. / Journal of Neuroscience Methods 259 (2016) 57–71

Table 1 Behavioral measures. Basic measures

Definition

Human PVT parameter (Basner and Dinges, 2011)

Premature responses

Responses occurring (1) prior to stimulus onset, or (2) within 150 ms after stimulus onset

Similar to false starts/errors of commission. These trials include responses without a stimulus or responses within 100 ms of stimulus onset

Correct responses

Responses occurring between 150 and 1500 ms after stimulus onset

A response occurring after first 100 ms of stimulus onset

Misses

No responses within 1500 ms after stimulus onset. Also included in lapses in attention (see below)

Similar to time out. A trial on which no response is made within 30 s of stimulus onset; the trial times out, but is counted as a lapse with an RT = 30 s (see Lapses below)

Reaction time

The elapsed time in milliseconds from stimulus onset to the occurrence of a correct response. Also referred to as response latency

The elapsed time in milliseconds from stimulus onset to the occurrence of a response. Also referred to as response time

Post-hoc measures

Definition

Human PVT parameter (Basner and Dinges, 2011)

Lapses

Number of Correct Responses with a latency > twice the mean latency for an animal’s entire session, plus the number of Misses

Responses with an RT > 500 ms and including time out trials described above (see Misses); also termed errors of omission

False alarm estimates

Subsets of premature responses defined as responses occurring (1) within the 3–10 s range of stimulus onset times (shaded area of Fig. 1), or (2) within 150 ms after stimulus onset. Together, 1 and 2 are used as an estimate of false alarm responding for signal detection analysis (see Section 2). Used primarily as a means to compare performance between rodent subjects and between published studies

Similar to false starts (see premature responses), but not a direct counterpart

measures that are acquired after the behavioral session following determination of each rat’s mean RT, for example. 2.4.1. Basic measures Reaction times (RT; i.e., response latencies or response times) were defined as the elapsed time in milliseconds from stimulus onset to the occurrence of a response; to count as a valid RT, this response needed to occur after the stimulus light was illuminated for 150 ms (see below) and before the end of the 1500 ms response window. Summary reaction time measures calculated included the fastest 10% of RTs (or Q-10), median RT (or Q-50), slowest 10% of RTs (or Q-90), and the mean RT; percentiles were employed in a manner similar to the human PVT since reaction time distributions are often skewed due to the physiological limits on reaction times (Stebbins and Miller, 1964). Premature responses were defined as responses prior to stimulus onset, as well as responses within 150 ms after stimulus onset, and likely not true responses to the stimulus since there is a physiological limit to the minimal speed with which an organism can respond (e.g., about 140–160 ms for humans). This measure is similar to false starts in the human PVT, where false starts are defined as responses prior to or within 100 ms of stimulus onset (see Table 1). In turn, correct responses were defined as responses occurring within the interval from 150 to 1500 ms after the light onset (thus excluding any responses with a reaction time of less than 150 ms). While correct responses are not a frequently reported outcome metric in the human PVT literature, correct responding is commonly reported for various rodent behavioral tasks and is reported here to enable comparison of the current rPVT to similar behavioral tests in the literature. Misses were defined as the number of trials with no responses occurring within 1500 ms after the stimulus onset. This outcome metric is most like “time out” trials in the human PVT, where the subject does not register a response within 30 s after stimulus onset. These trials, however, are considered valid trials in the human PVT and are recorded as lapses with an RT of 30 s. Thus, misses in the rodent version of the PVT are included in the calculation of lapses, but the RT values for missed trials are not included in any RT calculation since no response was emitted (see Table 1). 2.4.2. Additional computed measures The primary measure of attention in the human PVT literature is the degree to which lapses in attention occur during the procedure. In the human PVT, lapses are considered to be the sum

of the number of time out trials (i.e., trials not responded to and resulting in a 30 s RT) plus those trials responded to with reaction times >500 ms (or >500 ms, depending on the study parameters; see Basner and Dinges, 2011). In the present report, we followed a definition of lapses similar to Christie et al.’s rPVT (Christie et al., 2008a, 2008b; Deurveilher et al., 2015; Oonk et al., 2015), where lapses were compared to each rat’s mean RT, instead of comparing them to a threshold value, due to the variability in RTs from individual rats. That is, lapses were defined as correct responses in which the RT was greater than twice the mean reaction time for each animal’s session plus omitted trials (i.e., misses, see Table 1). Lapses were calculated after the completion of each session by first computing the mean RT during each session, and then using that session mean to calculate lapses during that session. Additionally, to provide a method for comparing behavioral performances across different rPVT rodent studies and for comparing across rodent rPVT and human PVT studies, subsets of premature responses were used to estimate false alarm rates (see Table 1). In animal SRT studies, false alarm rates are measured by including a small percentage of trials during which no stimulus is actually presented, and measuring the frequency with which responding occurs on these trials (i.e., false alarm = response in the absence of the stimulus). While no false alarm trials occurred in the present study, a false alarm estimate was derived by determining the percentage of premature responses that occurred during the actual 3–10 s interval when a stimulus could have appeared (shaded area in Fig. 1). Using premature responses to estimate a false alarm rate allowed for the calculation of a d index of signal discriminability in which rates of percent correct (PC) responding and false alarm (FA) responding were converted into z scores, and subtracted (d = z(PC) − z(FA); Macmillan and Creelman, 1991). This d estimate was then employed for comparisons of the present data with other rodent data, and with human PVT performances. Total trials completed were defined as the number of premature responses plus correct responses plus misses. Given that that the session durations of the human and rat PVTs differ (typically 10 min and 30 min, respectively), rats’ performances are presented here as a percentage of total trials, instead of the total number of each outcome metric (e.g., percentage of trials on which a lapse occurred instead of total number of lapses). Further, presenting performance data as percentages also allows comparisons between other rodent studies as well. Baseline performances were defined as stable when the following conditions were met: the percentage

C.M. Davis et al. / Journal of Neuroscience Methods 259 (2016) 57–71

of correct responses was 75% or greater during a session; percentage of premature responses were less than 30% for a session; and there were no systematic changes in these measures or in median reaction times across successive sessions. 2.4.3. Time on task and the vigilance decrement Vigilance is defined as the ‘ability of a subject to maintain its focus of attention and to remain alert to stimuli over prolonged periods of time’ (Warm et al., 2008), and as the amount of time performing a detection task increases, vigilance decreases. This decline in vigilance over time spent on a task is known as the vigilance decrement and can appear within 5 min of task onset, depending on the demands of a given vigilance task. The vigilance decrement can be assessed on the human PVT by examining the change in mean 1/RT (or speed) as a function of time performing the task. Non-fatigued subjects typically display a decrease in speed (i.e., decreased performance) with increasing time on task, and this decrease occurs more rapidly and to a greater extent following varying levels of sleep deprivation (Lim et al., 2010; Raymann and Van Someren, 2007). In humans, time on task is usually assessed by examining performance during consecutive time intervals during a PVT session, such as average performance at each minute on the 10-min PVT. Given that this is an important measure of sustained attention on the PVT, time on task data for rats performing the rPVT are presented below for speed (i.e., mean 1/RT), as well as for other performance metrics (e.g., correct responses, premature responses, mean RT, median RT, etc.). For time on task calculations in the current study, the 30 min rPVT session was divided into 5 bins: (1) 0–6 min, (2) 6.1–12 min, (3) 12.1–18 min, (4) 18.1–24 min, and (5) 24.1–30 min. In this way, comparisons between the vigilance decrement on the human PVT and rPVT can be made. 2.4.4. Variable response-stimulus interval effect (foreperiod analyses) Tucker et al. (2009) demonstrated that humans performing the PVT displayed longer RTs and greater numbers of lapses when the immediately preceding response-stimulus interval (RSI) was short. In contrast, RTs were shorter at longer RSI’s, but false start responses increased. Interestingly, these RSI relationships were not related to the same subjects’ time on task performance, were unaffected by sleep deprivation, and were argued to be controlled by a different neural mechanism (Tucker et al., 2009). In order to better characterize rats’ performances during the rPVT, analyses of the various performance metrics were also assessed as a function of the variable RSI (i.e., the variable foreperiod). Since these foreperiod values occur randomly across the session in the rPVT in a manner similar to that in the human PVT, performances can be specific to the foreperiod duration (i.e., the amount of time an animal is required to wait for stimulus onset). For the current study, performance measures were averaged for each foreperiod value; for example, the 4 s foreperiod value consists of average performance from all foreperiods between 3 and 4 s; the 5 s foreperiod value consists of average performances from foreperiods between 4.1 and 5 s, etc. This analysis pattern resulted in 7 foreperiod bins (see below), with on average, about 30 trials occurring in each bin across a typical session of 205 trials. Given the discrepancy between the RSI effect on the human and rat versions of the PVT, we performed an RSI analysis (i.e., foreperiod analysis) to determine if the parameters used in the current version of the rPVT would lead to an RSI effect similar to that reported in humans. 2.5. Experimental interventions 2.5.1. Drug administrations The effects of acute injections of d-amphetamine (0.056–1.0 mg/kg) or zolpidem (0.3–3.0 mg/kg) were examined by injecting rats (n = 4) two times per week, typically Tuesday and

61

Friday. Since both compounds are short acting (half-lives of 10 and 3 h, respectively), this schedule minimized any possible carryover effects between successive injections. Drugs were diluted in 0.9% NaCl with the total concentration adjusted to yield the appropriate dose at 1 ml/kg volume. During sessions when a drug was not administered, 1 ml/kg 0.9% NaCl was injected. Drug and saline were administered via intraperitoneal (i.p.) injections prior to placing an animal in a test chamber (10 min pretreatment for amphetamine; 30 min pretreatment for zolpidem) and were given at approximately the same time every day. Doses were given in a random order and each dose was administered at least twice in each animal. Before drug injections could be given, baseline responding on the 3 days prior to a drug injection were required to meet the following criteria: (1) the percentage of correct responses was 75% or greater during a session, (2) false-alarm rates were less than 30% for a session, and (3) there were no systematic changes in the time course of these measures or in median reaction times across successive sessions. 2.6. Statistical analyses Behavioral parameters during acquisition performance were assessed with repeated-measures ANOVAs, with the repeated factor of either Session, Time on Task Value, or Foreperiod Value. FDR-corrected paired t-tests were used as post-hocs to assess differences between Sessions. Tukey-corrected post-hocs were used to assess Time on Task or Foreperiod Value differences between Weeks. In the Time on Task or Foreperiod repeated-measures ANOVA, Week was treated as a between-subjects factor. Pairedsamples t-tests with an FDR correction were used to assess within week differences. Behavioral data following drug injections was analyzed for each rPVT performance measure and drug separately using repeated-measures ANOVAs, where Foreperiod Value was the repeated factor and Drug Dose was the between-subjects factor. Tukey-corrected post-hocs were used to assess specific group differences. All statistical analyses were performed with the Statistical Package for the Social Sciences (SPSS, v20.0). 3. Results 3.1. Acquisition of performance Performances under the rPVT procedure were acquired quickly, with the average number of automated training sessions needed to reach the final performance parameters on the rPVT procedure being 9.3 sessions (SD = 3.2; N = 122). During the first full session under the final performance parameters (i.e., Session 1; see Fig. 2), rats averaged 73.4% correct responding, 18.6% lapses, 10.6% premature responding, and 527 ms reaction times. The d index of discriminability averaged 2.36 on Session 1. Fig. 2 shows the percentages of correct responses, lapses in attention, and premature responses, respectively for all rats performing the rPVT. Fig. 2 shows that, starting with Session 1 all performance measures were indeed fairly stable from the start and closely approximated the steady-state levels of these measures that were achieved within the next 10–12 sessions. As shown in the figure, performance levels remained extremely stable for the next 80 sessions. The repeated-measures ANOVA revealed a significant within-subjects main-effect of Session [F(19, 2052) = 7.244, p < 0.05]. Correct responding increased slightly, but non-significantly, over Sessions 1–3 (Session 1 compared to Session 2 or 3, all p’s > 0.203). By Session 4, this increase in correct responding was significant when compared to Session 1 (p = 0.002) and remained stable thereafter, with no significant differences between consecutive sessions from Session 4–10. To demonstrate stability of correct responding, correct responding for Session 10 was compared to each 10th

62

C.M. Davis et al. / Journal of Neuroscience Methods 259 (2016) 57–71

Fig. 3. The d signal detectability index for rats (n = 122) over the first 80 sessions of performing the rPVT. Error bars = ±SEM.

Fig. 2. Percentages of correct responding, lapses in attention, and premature responding for rats (n = 122) performing the rPVT. Misses are included on the graph in order to compare this measure with lapses, and to show that more than half of the lapses during normal rPVT performances were misses and the remainder were correct trials on which the subject responded with an RT two times (or greater than) the mean RT for that session. Performance scores for Session no. 1 represent performances for the first full session under the final performance parameters. Error bars = ±SEM.

session (e.g., Session 20, 30, etc.) up to Session 80. The only significant difference occurred between Session 10 and Session 70 (p = 0.007), where correct responding on Session 10 was 78.63% compared to 83.8% on Session 70. Thus, rPVT performances were

stable over time, with minimal increases in correct responding over 80 sessions (approximately 4 months). A similar change was evident for d values, where the repeated-measures ANOVA revealed a significant within-subjects main effects of Session [F(19, 1862) = 2.986, p < 0.05]. While there was a trend for average d to increase over approximately the first 20 sessions, these increases were not significantly different when individual sessions were compared to one another, e.g., session 1 compared to session 4. In a manner similar to the analysis of correct responding above, d for Session 10 was compared to each 10th session up to Session 80. Session 10 was significantly different from Session 30 (p = 0.002; d = 2.34 and 2.63, respectively) and Session 70 (p < 0.05; d = 2.79); Session 60 also differed from Session 70 (p = 0.001; d for Session 60 = 2.42). Fig. 3 shows the corresponding stability of the d index

Fig. 4. Comparisons of the percentages of premature responding, correct responding, lapses in attention, d index, and response latencies for rats (n = 92) performing the rPVT as compared to those of Christie et al. (2008a, 2008b), Loomis et al. (2015), and Oonk et al. (2015). Error bars = ±SEM. No error estimates were available for correct responding from the Christie et al. (2008b) study.

C.M. Davis et al. / Journal of Neuroscience Methods 259 (2016) 57–71

63

and females in the general SRT literature (Adam, 1999; Barral and Debu, 2004; Bellis, 1933; Botwinick and Thompson, 1966; Dane and Erzurumluoglu, 2003; Engel et al., 1972; Noble et al., 1964; Welford, 1980). The median RTs for the male and female rats of Fig. 5 were quite similar, being 410 and 404 ms, respectively. Fig. 6 displays comparisons of premature responding, lapses in attention, correct responding, and the d signal detectability index for human adolescents performing the PVT (Beijamini et al., 2008) and for rats performing the current version of the rPVT. This study was chosen as a comparison because it included percentage scores of several performance metrics, from which percent correct responding was derived and plotted in Fig. 6. Measures of premature responding, lapses in attention, correct responding, and signal detectability were all quite similar between the boys and girls of the Beijamini et al. study and the rats of the present study. Additionally, the variability in these measures was also quite comparable for humans and rats. Fig. 5. A comparison of the latency distributions observed with men and women (redrawn from Blatter et al., 2006) as contrasted with both male and female rats performing the rPVT.

of discriminability over the first 80 sessions, and a slight increase in discriminability over this period described above. Stability of baseline performances was also evaluated by determining the percentage of animals that reached the stability criteria employed in a previous study (i.e., percent correct responding >75% and premature responding 0.067), reaction times for the first three intervals for Week 1 differed from all other weeks except Week 2 (all p’s < 0.045). Q-90 reaction times decreased for these first three intervals as training progressed, and thus, the pattern of Q-90 reaction times across time on task interval was stable by Week 3. Within-week analyses for Weeks 5–20 again demonstrated that for most performance variables, the first time on task interval differed from the last time on task interval. Specifically, premature responses were significantly greater during the first time on task interval compared to the last interval for each week from Weeks 5–20, such that the pattern for Week 5 illustrated in Fig. 8 was maintained throughout the 15-week period. Lapses, Q10, Q50 and Q90 reaction time measures were all significantly lower or faster, respectively, during the first interval compared to the last interval for each week from Weeks 5 to 20 (all p’s < 0.001), with this pattern being consistent across the remaining weeks. Percent correct responding was lower during the first interval compared to the last interval for weeks 5–9 (all p’s < 0.041), but from week 10 to 20, these intervals did not differ, demonstrating that percent correct responding was the same across intervals during these later weeks. Taken together, these data demonstrate that rats’ within-session performances on the rPVT are sensitive to a vigilance decrement (i.e., rats’ performances are slower and/or slightly degraded as a function of

elapsed session time in a manner similar to humans performing the PVT). 3.5. Variable foreperiod effects (response-stimulus interval analyses) Behavioral performances on the rPVT can also vary as a function of the variable foreperiod (i.e., the length of time an animal is required to wait before the light signal comes on during each trial, sometimes referred to as the response-stimulus interval, or RSI, in the human PVT). Fig. 9 shows the average weekly performances of rats for percent correct responding, premature responding, lapses in attention, the fastest (Q-10) reaction times, median (Q50), and slowest (Q-90) reaction times as a function of foreperiod duration or “wait time” in the rPVT. For correct responding, the ANOVA revealed only a main-effect of Week [F(19, 1558) = 8.856, p < 0.05], but no Foreperiod × Week interaction (p = 0.980). Tukeycorrected post-hocs demonstrated that only Week 1 differed from all Weeks 2–10 (all p’s < 0.0004); Week 2 differed from Weeks 6, 7, 9, and 13–15 (all p’s < 0.013). Starting at Week 3, there were no differences between any of the remaining 17 Weeks (all p’s > 0.201). Thus, by Week 3, correct responding across foreperiod duration was stable; these findings mirror the daily Session data (Fig. 2), where mean correct responding for each session stabilized between 4 and 10 sessions (approximately one to two weeks). For premature responding, the ANOVA revealed similar effects with a Foreperiod × Week interaction [F(114, 8550) = 1.722, p < 0.001]. As seen in Fig. 9, premature responding was significantly greater at the lowest foreperiod values (i.e., 4–6 s) during

66

C.M. Davis et al. / Journal of Neuroscience Methods 259 (2016) 57–71

Fig. 9. Average weekly performances of rats (n = 122) for percent correct responding, premature responding, lapses in attention, the fastest (Q-10) reaction times, median (Q-50) reaction times, and the slowest (Q-90) reaction times. Data are shown as a function of the variable foreperiod duration, or “wait time” required prior to the stimulus light coming on. For percent correct responding, Week 1 was on average significantly lower than all other weeks, but the pattern of responses was similar across foreperiod values; the same difference was apparent for Q-90 reaction times. Asterisks (*) denote a significant difference between Week 1 and all other weeks at specific foreperiod values (p < 0.05). While lapses significantly decreased as foreperiod value increased, that difference was consistent across all weeks. Error bars = ±SEM.

Week 1 compared to these same values at all other weeks (all p’s < 0.001), but significantly decreased over Weeks 2 and 3 (all p’s < 0.001), but remained significantly greater than premature responding at Weeks 11–20; no differences were apparent for this foreperiod value for Weeks 2 or 3 compared to Weeks 4–10. No differences were found at Week 4 in premature responding at each foreperiod value when compared to each value during Weeks 5–20 (all p’s > 0.107); thus premature responding across Weeks 4–20 was stable and is accurately represented by the Week 5 curve in Fig. 9. Lapses at each foreperiod value were similar across Weeks (p = 0.828), which demonstrates that the pattern of lapses emitted across foreperiod values was relatively stable across the various weeks of training; this stability can be see in Fig. 9. For the various reaction time measures, there were more significant changes across weeks with Q-10 for the various foreperiods than for both Q-50 and Q-90. The Foreperiod × Week interaction [F(108, 8502) = 3.497, p < 0.001] and Tukey-corrected post-hocs for Q-10 reaction times show that these RTs were the slowest during Week 1 (p’s < 0.012 for all foreperiod values except the 10 s foreperiod where no differences were found) and became significantly faster across weeks, until a stable pattern was reached by Week 3 (all p’s at all foreperiod values >0.109), with the shapes of the Week 3 and 5 curves shown in Fig. 9 representative of the stable Q-10 RTs across foreperiod values. For Q-50 RTs, a significant Foreperiod × Week interaction [F(108, 8508) = 1.444, p = 0.002] and Tukey-corrected post-hocs showed that Q-50 values were slower at lower foreperiods than longer ones, with these RTs being significantly slower for the 4–9 s foreperiod values during Week 1 compared to all weeks except Week 2 (all p’s < 0.05). Week 1’s Q-90 RTs were slower on average, but the pattern of Q-90 RTs by foreperiod value did not differ across weeks, which was demonstrated by the lack of a significant Foreperiod × Week interaction (p = 0.683),

but significant main-effect of Week [F(18, 1417) = 5.774, p < 0.001]. On average, Weeks 1 and 2 were slower than all other weeks (all p’s < 0.028), and no significant changes were apparent in the pattern of Q-90 RTs after Week 4. In general, these foreperiod effects demonstrated that, as an animal is required to wait longer during the foreperiod interval, percent correct responding decreases and premature responding increases, indicating a decrease in stimulus discriminability. Importantly, as the foreperiods vary randomly throughout a session, this decrease in stimulus discriminability is a function of waiting for longer, but not shorter, foreperiods. This finding is in opposition to the time on task effect, i.e., a decrease in task performance at later times within a session due to a vigilance decrement, irrespective of the variable foreperiod for each trial, which lends support from an animal model for Tucker et al.’s (2009) hypothesis that the RSI effect and time on task have different neural mechanisms; this hypothesis could be investigated using this version of the rPVT. Additionally, lapses in attention decrease markedly at longer foreperiods. Finally, all response latencies are longest when shorter foreperiods occur, and shorten considerably at the longer foreperiod lengths. 3.6. Drug effects on the rPVT Fig. 10 shows dose-effect functions for correct responding, premature responding, lapses in attention, and response speed following administration of doses of d-amphetamine (0.56–1.0 mg/kg), zolpidem (0.1–3.0 mg/kg), or vehicle. Only the high dose of zolpidem produced pronounced effects on mean performances by lowering accuracy and response speed, and elevating lapses in performances. A slightly different picture emerged when these same data were subjected to a more fine-grain analysis by looking at mean performance effects at the different

C.M. Davis et al. / Journal of Neuroscience Methods 259 (2016) 57–71

67

Fig. 10. Session averages (n = 4) for correct responding, premature responding, lapses, and mean speed following i.p. administration of vehicle, d-amphetamine (0.56–1.0 mg/kg), or zolpidem (0.3–3.0 mg/kg). The highest dose of zolpidem significantly decreased mean correct responding, and increased both mean lapses and mean speed. Asterisks (*) denote significant differences from vehicle; p < 0.05. Error bars = ±SEM.

foreperiod durations (Fig. 11). A repeated-measures ANOVA assessing amphetamine’s effects on correct responding revealed a significant within-subjects effect of Foreperiod [F(6, 108) = 13.3, p < 0.001] and significant within-subjects interaction of Foreperiod × Dose [F(30, 108) = 2.027, p = 0.004]. The between-subjects main-effect of Dose was not significant (p = 0.721; this finding supports the lack of changes in mean session performance by amphetamine illustrated in Fig. 10). Tukey-corrected post-hocs revealed that amphetamine’s effects were specific to the 4-s foreperiod duration where 1.0 mg/kg amphetamine significantly increased accuracy compared to vehicle, 0.056, and 0.1 mg/kg amphetamine (Fig. 11; for clarity, 0.056 and 0.1 mg/kg data not shown). Although the 0.32 and 0.56 mg/kg doses increased accuracy at this foreperiod, these effects were not significant (all p’s > 0.131; data not shown). These improvements, however, were not apparent at the other foreperiod durations. This lack of effect is most likely due to the fact that accuracy values following vehicle administration were already very high at these longer durations. A similar foreperiod-dependent effect was apparent for lapses, as well, with a significant within-subjects effect of Foreperiod [F(6,108) = 56.944, p < 0.001] and significant within-subjects interaction of Foreperiod × Dose [F(30,108) = 1.886, p = 0.01]. In a manner similar to the data for percent correct responding, 1.0 mg/kg dose of d-amphetamine significantly reduced lapses at the 4- and 5-s foreperiod durations, but not at the longer durations, when compared to vehicle. The within-subjects effect of Foreperiod was significant for premature responding [F(6, 108) = 11.235, p < 0.05] and mean reaction times [F(6, 108) = 6.051, p < 0.001], but no interaction with amphetamine dose (all p’s > 0.12), which suggests that the typical pattern of increased premature responses at

longer foreperiod values was not affected by amphetamine administration. For zolpidem, a foreperiod analysis on percent correct responding showed significant within-subjects main-effects of foreperiod [F(6, 90) = 11.225, p < 0.001] in the same manner as that reported above for amphetamine, but no within-subjects interaction with zolpidem dose (p = 0.521). However, betweensubjects main-effects of zolpidem dose were significant [F(4, 15) = 6.706, p = 0.003], suggesting that zolpidem’s effects differed by dose, but were consistent across foreperiod duration. The highest dose of zolpidem, 3.0 mg/kg, significantly decreased percent correct responding compared to vehicle and all other zolpidem doses 0.1–1.8 mg/kg; all p’s < 0.014). No other dose induced a significant change in percent correct responding. For lapses, a significant within-subjects effect of Foreperiod [F(6, 90) = 34.312, p < 0.001] was found, but no interaction with zolpidem dose (p = 0.953); the between-subjects effect of zolpidem dose was significant [F(4, 15) = 10.134, p = 0.0003]. Again, the 3.0 mg/kg dose significantly increased lapses compared to vehicle and all other doses of zolpidem. For mean reaction times, a significant within-subjects effect for Foreperiod [F(6, 90) = 10.777, p < 0.001] was found, but no significant interaction with zolpidem dose (p = 0.193). The highest dose of zolpidem significantly increased mean speed on the rPVT compared to vehicle and the lowest dose (0.3 mg/kg); this increase was not different from the increases induced by the 1.0 and 1.8 mg/kg doses of zolpidem. While premature responding, on average, decreased as the dose of zolpidem increased, these effects were not significant (all p’s > 0.346), nor did premature responding differ across the foreperiod durations (p = 0.545).

Fig. 11. Mean performances (n = 4) on each foreperiod value of the rPVT for percent correct responding, premature responding, lapses, and mean speed following i.p. administration of vehicle, d-amphetamine (1.0 mg/kg), or zolpidem (3.0 mg/kg). d-Amphetamine significantly increased correct responding and decreased lapses at the shortest foreperiod durations. Asterisks (*) denote significant difference from vehicle; p < 0.05. Zolpidem significantly decreased correct responding, increased lapses, and slowed mean speed; these effects were consistent across foreperiod value. For lapses, small data points indicate the proportion of lapses from missed trials (“errors of omission”) only, and indicate that the elevation in lapses following zolpidem administration was strictly due to missed trials. Error bars = ±SEM.

68

C.M. Davis et al. / Journal of Neuroscience Methods 259 (2016) 57–71

Fig. 12. Mean speed as a function of time on task in the rPVT following i.p. administration of doses of vehicle, d-amphetamine (1.0 mg/kg), and zolpidem (3.0 mg/kg). d-Amphetamine slightly, but non-significantly, increased speed throughout sessions, while zolpidem slowed response speed early in the session. Asterisks (*) denote significant difference from vehicle. Error bars = ±SEM.

Fig. 12 shows that zolpidem also significantly changed the time on task performances. Fig. 12 shows mean speed plotted as a function of time on task in the rPVT following i.p. administration of doses of vehicle, d-amphetamine (1.0 mg/kg), and zolpidem (3.0 mg/kg). d-Amphetamine produced small, nonsignificant (p = 0.910) increases in response speed throughout the session, relative to performances following vehicle. Both time on task curves, however, still show a gradual slowing in response speed across the session (within-subjects main-effect of Time on Task: F(4, 72) = 5.575, p = 0.001). Only the 3.0 mg/kg dose of zolpidem, however, dramatically slowed response speed throughout the session [Time on Task × Dose interaction: F(16, 60) = 1.894, p = 0.039]. At the three earliest time on task intervals, response speed was significantly slower following 3.0 mg/kg zolpidem when compared to vehicle (following Tukey-corrected post-hocs, all p’s < 0.041). There was a slight recovery in response speed toward the end of the session, such that no difference between zolpidem and vehicle was found at the last two time on task intervals (all p’s > 0.316); this effect could be due to a lessening of the sedativelike effects of the drug over time. 4. Discussion The results of the present study show clearly that the present rPVT produces performances in rats that are comparable in many respects to human PVT performances. Rats quickly learn the rPVT and show a clear discrimination that is characterized by high rates of percent correct responding and low rates of premature responding. Additionally, the frequency distributions of rats’ reaction times to stimulus onset are quite like those produced by humans (Blatter et al., 2006), and rates of both premature responding and lapses are also similar to those reported for human adolescents in the literature (e.g., Beijamini et al., 2008). Finally, the time-ontask and inter-stimulus interval effects often reported for humans can also be seen in the rodent performances. This approximation of human performances on the human PVT represents a distinct advantage of the rPVT as a pre-clinical platform for translational research targeting human vigilance and neurobehavioral function. 4.1. rPVT performance comparisons Performances of rats under the present rPVT were more comparable to human PVT performances than those reported by Christie

et al. (2008a, 2008b) and other investigators employing those procedure parameters (Loomis et al., 2015; Oonk et al., 2015) with major differences being in the lower percent correct responding and higher rates of premature responding reported in the these studies. Numerous differences exist between the present study and those of the other cited rodent studies including the rat strains employed (Long-Evans vs. Fischer vs. Wistar), the type of reinforcer (water vs. food pellets) used, and the parameters used to establish the basic performances. The present procedure, however, has also been used with both Fischer 344 and Lewis strains of rats in this laboratory, with little differences noted in their performances compared to those observed with the Long-Evans strain (Davis et al., 2015). While Oonk et al. (2015) used male Long-Evans rats as subjects, performances were similar to those in the Christie et al. studies, and not the Long-Evans rats’ used in the current study, again suggesting that the differential task parameters are the underlying mechanism of improved performances in the current study. Loomis et al. (2015) employed male Wistar rats, food pellets as reinforcers, and a specific training regimen to minimize premature responding; performances of their rats were most similar to those observed with the present procedure. Differences in the type and frequency of reinforcement employed in other SRT procedures indicate that response latencies can vary as an inverse function of reinforcement frequency and reinforcement magnitude (Stebbins, 1962; Stebbins and Lanson, 1962). Thus response latencies are longer under intermittent schedules of reinforcement, as well as when the reinforcer magnitude (i.e., percent sucrose solution delivered) is decreased. However, none of the above rPVT studies nor the present study varied reinforcement parameters in this manner. The differences between the prior rPVT studies and the present study are more likely related to the differing parameter values employed, with prior studies typically using a variable 3–7 s response-stimulus interval (i.e., foreperiod) and a correct response window of 2.5–3.0 s following stimulus onset, compared to the present study with a 3–10 s variable response-stimulus interval and a correct response window of 1.5 s following stimulus onset. Not only is the longer foreperiod of the present study more comparable to the response-stimulus interval suggested as optimum for most human PVT studies (Basner and Dinges, 2011), but also it greatly lowers the probability of an animal correctly responding on a trial by chance. For example, if a response is emitted 4 s after the end of the prior reinforcement, there would be a 75% chance of that response being correct, given a 3–7 s foreperiod and a 3-s response time criterion. With a 3–10 s foreperiod and a 1.5 s response time criterion, the chance of that same response being correct would be 21%. Thus there is a much greater chance of responding being randomly reinforced when shorter foreperiod ranges are employed, which results in a reduced amount of stimulus control (i.e., a greater proportion of responding in the absence of the stimulus or premature responses), which was the case in the published reports employing the Christie et al. parameters (Christie et al., 2008a, 2008b; Loomis et al., 2015; Oonk et al., 2015). A seeming exception to this in the human literature might be the use of the ‘brief’, 3-min PVT, or PVT-B (Basner et al., 2011) that employs a variable foreperiod of 1–4 s. However, when compared to the more standard 10-min PVT, the PVT-B produces shorter RTs, higher premature responding (errors of commission), and lower lapse frequencies, resulting in a decreased sensitivity of the PVT-B to detect sleep loss. Basner et al. countered this decreased sensitivity by reducing the typical human lapse threshold of 500 ms to 355 ms, a strategy that effectively lowers the probability of responses being categorized as correct. Walker et al. (2011) adapted the rPVT for use with restrained rats so that physiological recordings could be obtained at the same time. In this version, the stimulus being responded to was

C.M. Davis et al. / Journal of Neuroscience Methods 259 (2016) 57–71

mechanical stimulation of a whisker, and the response was licking a sipper tube to receive a sucrose solution as the reinforcer. False alarm rates were obtained only during single baseline sessions when the whisker was either not placed in the stimulator, or the stimulator was not active. With a varying ITI of 8 to 16 s, this study reported percent correct responding and false alarm responding of approximately 93% and 33%, respectively, which represents a d value of 1.92 compared to the present study’s average d value of 2.8. Direct stimulation of the whiskers in the Walker et al study produced mean RTs of about 195 ms, a value considerably faster than those observed in the present study; mean RTs lengthened following 12 h of sleep deprivation, but not after 3 or 6 h of sleep deprivation. Data on lapses or premature responding were not reported, however, making it difficult to judge the degree to which this variation of the rPVT procedure mimics human vigilance performances. The use of direct stimulation of the whiskers as the “vigilance stimulus”, however, does make this procedure quite different from a typical human PVT since it provides an immediate alerting stimulus, regardless of whether the animal may be visually attending to its surroundings or not. In general, differences between the rodent rPVT and the human PVT exist due to the need for proper contingencies of reinforcement to shape and maintain the rPVT performance in rats, as opposed to the sole use of instructional control for obtaining good PVT performances in humans. Differences in mean reaction times between humans and rats under the PVT and rPVT procedures are likely an effect attributable to these contingency differences as well as topographical differences in responding (e.g., human finger tap vs. rodent nose-poke or lever press). Human RTs can be up to 50% shorter than those of adult rats, which is likely due to the typical instructions given to humans first performing the PVT that stress the need to attend to the display and produce as quick a reaction time as possible. While such instructional control can be translated into explicit contingencies of reinforcement in animal experiments to produce faster reaction times (e.g., by reinforcing only responses that meet a specific latency requirement; see Moody, 1970; Pfingst et al., 1975; Stebbins, 1966; Stebbins and Miller, 1964), imposing these contingencies can considerably lengthen the training time required to produce fast yet stable RTs (Hienz and Weerts, 2009; May et al., 1995; Pfingst et al., 1975). In a number of SRT studies, the control needed to signal an animal to “pay attention” has been accomplished via the use of an “observing response” that must be executed by the animal before the reaction time stimulus is presented (Hienz et al., 2008; Pfingst et al., 1975; Stebbins, 1966; Stebbins and Miller, 1964; Weed et al., 1999). The use of such observing responses insures that the animal is “ready”, or in a position to quickly respond with maximum speed when the stimulus is presented and thus minimizes the occurrence of missed trials since a reaction-time stimulus is never presented when an animal has not already made the observing response, and thus is not ready to respond with a relatively quick reaction time. However, the use of an observing response necessarily reduces the effectiveness of the procedure for detecting lapses in attention – a critical variable throughout the human PVT literature – since lapses in attention are extremely unlikely to occur once an observing response has been initiated. Further, as noted above, the RT values obtained with the present rPVT procedure are not that exceedingly different from those observed in human PVT studies. A major difference between human and animal PVT performances is the fact that most animal behavioral studies necessarily use positive reinforcement (food, water, sucrose, etc.) to maintain the relevant behavior over relatively long periods, as was the case in the present study. Under these conditions, the time on task effects typically seen in humans within 10 min PVT sessions occurred over a longer interval (i.e., 30 min) in the present study

69

(e.g., Fig. 4, left and right panels), which were likely due to the maintenance of attention via food reinforcement, and represents a difference to be aware of when comparing human and non-human PVT performances. 4.2. Time on task effects in the rPVT The parameters used in the rPVT procedure resulted in time on task effects that were primarily restricted to changes in RTs and lapses. For example, once stable (i.e., data from Week 5 and later), rats’ performances were better at the early time points in the sessions, showing lower percentage of lapses and shorter median and Q-90 RTs compared to later time points. These results indicate that a time on task vigilance decrement can be demonstrated in healthy, non-sleep deprived rats performing the rPVT. Further, the time on task RTs increased approximately 100 ms in the absence of a decrease in percent correct responding, suggesting that even though rats’ reaction times were slowing within the sessions, they were still responding correctly throughout the entire session. While it could be argued that the slowing of responding could be due to satiation from receiving food throughout the session, an increase in omissions (i.e., decreased correct responding) would also be expected, but this increase did not occur. Further, given that lapses increased in the absence of changes in correct responding, the increase in lapses was not simply due to an increase in misses or omitted trials. However, rodent performances do need to stabilize prior to using time on task data for assessing within-session changes in vigilance, as can be seen from the significant differences in the time on task data between the early weeks of training (Weeks 1–5; Figs. 7 and 8). 4.3. Foreperiod, or inter-stimulus interval effects in the rPVT When analyzing rats’ performances on the rPVT by foreperiod (inter-stimulus interval), once performances stabilized, a clear effect was evident such that performance measures changed as a function of foreperiod value. More specifically, at short foreperiod values RTs were longer, lapses were higher, and few, if any, premature responses occurred. At longer foreperiod values, however, RTs became shorter, premature responses increased, and lapses decreased. These patterns displayed by rats on the rPVT are very similar to the response-stimulus interval (RSI) effect seen in humans performing the PVT as described by Tucker et al. (2009). Further, Oonk et al. (2015) looked for an RSI effect in rats performing the Christie et al. version of the rPVT, but did not find it; from their data, Oonk et al. argued for improvement in ‘several rPVT parameters’ to increase the rodent version of the test’s experimental validity. The present data indicates that the RSI effect can be demonstrated in rats performing the rPVT with the present parameters. 4.4. Drug effects on the rPVT Drug administrations had predictable effects on rPVT performance measures, and in a manner similar to the human PVT and other animal tasks assessing sustained attention. In humans, psychostimulants such as d-amphetamine and d-methamphetamine have been shown to improve vigilance performance (e.g., decrease lapses, increase correct trials, decrease omissions) on several tasks, including the PVT (Bonnet et al., 2005; Caldwell and Caldwell, 1997; Caldwell et al., 2003; Cochran et al., 1992; Hartmann et al., 1977; Kenagy et al., 2004; Killgore et al., 2008; Lim and Dinges, 2008b; Magill et al., 2003; Silber et al., 2006; Waters et al., 2003; Wesensten et al., 2005). While most of these human studies used sleep-deprived subjects, when assessed in non-fatigued subjects, d-amphetamine still improved vigilance performance in

70

C.M. Davis et al. / Journal of Neuroscience Methods 259 (2016) 57–71

both humans and rats (Evenden et al., 1993; Loomis et al., 2013, 2015; Silber et al., 2006). In a manner similar to these previous reports, amphetamine administration produced increases in percent correct responding and decreases in lapses. Interestingly, these effects on rPVT performance were only apparent for trials with shorter, compared to longer, foreperiod durations. This effect is most likely due to the fact that non-drug performances were lower at these shorter foreperiods compared to the other foreperiod durations (i.e., the RSI effect), and there was less of a ceiling effect for improvement following amphetamine administration at these shorter foreperiods. Thus, amphetamine dose-dependently improved performance based on the RSI effect, such that drug administration significantly decreased RTs and lapses at shorter foreperiods, but had no significant effect on performances at longer foreperiods; to the best of our knowledge, no human PVT data assessing the effects of amphetamine on the RSI effect exists in the published literature. While animal studies examining the RSI effect on choice RT tests have been published, none of these studies assessed the performance-altering effects of psychostimulants (Rabbitt, 1969; Wilkinson, 1990). Amphetamine, however, has been extensively examined in other attention tasks (Evenden et al., 1993; Loomis et al., 2013, 2015; Paterson et al., 2011) and an effect of amphetamine as a function of inter-trial interval (ITI; similar to the response-stimulus interval) has been reported for percent correct responding, premature responses, omissions, and perseverative responses on the five-choice serial reaction time task (5-CSRTT), where amphetamine’s effects varied as the ITI duration was increased, most likely due to the baseline behavior at each ITI value (Paterson et al., 2011). At 1.0 mg/kg, amphetamine had no effect on percent correct responding at the 4 s ITI in the 5-CSRTT, where this dose did improve performance at the 4 s foreperiod duration in the rPVT. This dose did increase premature responding at the 5, 7, and 10 s ITI’s on the 5-CSRTT, and while not significant, there was an increase in premature responding at the longer foreperiod values on the rPVT following this dose of amphetamine. More work is needed to compare these two tasks to determine how similar a variable ITI in the 5-CSRTT is to the variable foreperiod values in the rPVT. Interestingly, amphetamine had no effect on the time on task for rats performing the rPVT. This finding not only demonstrates the dissociation between time on task and the RSI effect in an animal model, but also supports the possibility of different neural mechanisms hypothesized for these effects (see Tucker et al., 2009). In contrast to the attention-improving effects of amphetamine, zolpidem caused a decrease in percent correct responding and an increase in lapses and mean reaction times. In support of zolpidem’s sedative effects, a more detailed analysis of the lapses following the 3.0 mg/kg dose revealed that these lapses consisted exclusively of response omissions (i.e., misses). Increases in reaction time and decreases in psychomotor performance, however, are commonly reported in humans following zolpidem administration (Berlin et al., 1993; Evans et al., 1990; Troy et al., 2000). 5. Conclusion In summary, the results show clearly that the present rPVT procedure can be employed to track changes in neurobehavioral performance following several types of independent manipulations. As with the human PVT, percent correct responding, lapses, and mean RTs can all be used to track performance differences resulting from the administration of drugs such as amphetamine and zolpidem. Additionally, the rPVT procedure clearly differentiates between control rats and those rats exposed to low levels of proton radiation (see Appendix A. Supplementary data). And the rPVT can also be used to document the effects of sleep deprivation, as has been shown in other rat

rPVT studies as well as the present one. Thus the rPVT can be an extremely useful tool in the investigation of the effects of a wide range of variables on a vigilance performance that compares favorably to the human PVT, may serve as an innovative translational platform for exploring the bases of individual vulnerability to neurobehavioral impairments, and for developing potential prophylactics, countermeasures, and treatments for neurobehavioral dysfunctions. Acknowledgements This research was supported by the National Space Biomedical Research Institute (NSBRI) through NASA cooperative agreement NCC 9-58, grants NBPF01604, NBPF02802, NBPF04201, PF02602 and EO00010, and NASA NNX15AC71G. Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jneumeth.2015. 11.014. References Adam JJ. Gender differences in choice reaction time: evidence for differential strategies. Ergonomics 1999;42:327–35. Ator NA. Subjects and instrumentation. In: Iversen H, Lattal KA, editors. Techniques in the behavioral and neural sciences: experimental analysis of behavior. Amsterdam: Elsevier; 1991. p. 1–62. Barral J, Debu B. Aiming in adults: sex and laterality effects. Laterality 2004;9:299–312 (Asymmetries of Body, Brain and Cognition). Basner M, Dinges DF. Maximizing sensitivity of the psychomotor vigilance test (PVT) to sleep loss. Sleep 2011;34:581–91. Basner M, Dinges DF, Mollicone D, Ecker A, Jones CW, Hyder EC, et al. 520-d mission simulation reveals protracted crew hypokinesis and alterations of sleep duration and timing. Proc Natl Acad Sci USA 2013;110:2635–40. Basner M, Mollicone D, Dinges DF. Validity and sensitivity of a brief psychomotor vigilance test (PVT-B) to total and partial sleep deprivation. Acta Astronaut 2011;69:949–59. Baunez C, Humby T, Eagle DM, Ryan LJ, Dunnett SB, Robbins TW. Effects of STN lesions on simple vs choice reaction time tasks in the rat: preserved motor readiness, but impaired response selection. Eur J Neurosci 2001;13:1609–16. Beijamini F, Silva AG, Peixoto CA, Louzada FM. Influence of gender on psychomotor vigilance task performance by adolescents. Braz J Med Biol Res 2008;41:734–8 (Revista brasileira de pesquisas medicas e biologicas/Sociedade Brasileira de Biofisica. [et al.]). Bellis CJ. Reaction time and chronological age. Exp Biol Med 1933;30:801–3. Berlin I, Warot D, Hergueta T, Molinier P, Bagot C, Puech AJ. Comparison of the effects of zolpidem and triazolam on memory functions, psychomotor performances, and postural sway in healthy subjects. J Clin Psychopharmacol 1993;13:100–6. Blatter K, Graw P, Munch M, Knoblauch V, Wirz-Justice A, Cajochen C. Gender and age differences in psychomotor vigilance performance under differential sleep pressure conditions. Behav Brain Res 2006;168:312–7. Bonnet MH, Balkin TJ, Dinges DF, Roehrs T, Rogers NL, Wesensten NJ. The use of stimulants to modify performance during sleep loss: a review by the sleep deprivation and Stimulant Task Force of the American Academy of Sleep Medicine. Sleep 2005;28:1163–87. Botwinick J, Thompson LW. Components of reaction time in relation to age and sex. J Genet Psychol 1966;108:175–83. Brown VJ, Robbins TW. Simple and choice reaction time performance following unilateral striatal dopamine depletion in the rat. Impaired motor readiness but preserved response preparation. Brain 1991;114(Pt 1B):513–25 (a journal of neurology). Caldwell JA, Caldwell JL. An in-flight investigation of the efficacy of dextroamphetamine for sustaining helicopter pilot performance. Aviat Space Environ Med 1997;68:1073–80. Caldwell JA, Caldwell JL, Darlington KK. Utility of dextroamphetamine for attenuating the impact of sleep deprivation in pilots. Aviat Space Environ Med 2003;74:1125–34. Cattell JMK. James McKeen Cattell, 1860–1944: man of science. Lancaster, PA: The Science Press; 1947. Christie MA, Bolortuya Y, Chen LC, McKenna JT, McCarley RW, Strecker RE. Microdialysis elevation of adenosine in the basal forebrain produces vigilance impairments in the rat psychomotor vigilance task. Sleep 2008a;31:1393–8. Christie MA, McKenna JT, Connolly NP, McCarley RW, Strecker RE. 24 hours of sleep deprivation in the rat increases sleepiness and decreases vigilance: introduction of the rat-psychomotor vigilance task. J Sleep Res 2008b;17:376–84.

C.M. Davis et al. / Journal of Neuroscience Methods 259 (2016) 57–71 Cochran JC, Thorne DR, Penetar DM, Newhouse PA. Parsing attentional components during a simple reaction time task using sleep deprivation and amphetamine intervention. Percept Mot Skills 1992;75:675–89. Dane S, Erzurumluoglu A. Sex and handedness differences in eye-hand visual reaction times in handball players. Int J Neurosci 2003;113:923–9. Davis CM, Decicco-Skinner KL, Hienz RD. Deficits in sustained attention and changes in dopaminergic protein levels following exposure to proton radiation are related to basal dopaminergic function. PLoS ONE 2015., http://dx.doi.org/10.1371/journal.pone.0144556 (in press). Davis CM, DeCicco-Skinner KL, Hienz RD. Individual differences in attentional deficits and dopaminergic protein levels following exposure to proton radiation. Radiat Res 2014;181:258–71. Deurveilher S, Bush JE, Rusak B, Eskes GA, Semba K. Psychomotor vigilance task performance during and following chronic sleep restriction in rats. Sleep 2015;38:515–28. Dinges DF, Orne MT, Whitehouse WG, Orne EC. Temporal placement of a nap for alertness: contributions of circadian phase and prior wakefulness. Sleep 1987;10:313–29. Dinges DF, Pack F, Williams K, Gillen KA, Powell JW, Ott GE, et al. Cumulative sleepiness, mood disturbance, and psychomotor vigilance performance decrements during a week of sleep restricted to 4–5 hours per night. Sleep 1997;20:267–77. Dinges DF, Powell JW. Microcomputer analyses of performance on a portable, simple visual RT task during sustained operations. Behav Res Methods Inst Comput 1985;17:652–5. Domenger D, Schwarting RK. The serial reaction time task in the rat: effects of D1 and D2 dopamine-receptor antagonists. Behav Brain Res 2006;175:212–22. Drummond SP, Bischoff-Grethe A, Dinges DF, Ayalon L, Mednick SC, Meloy MJ. The neural basis of the psychomotor vigilance task. Sleep 2005;28:1059–68. Eckart MT, Huelse-Matia MC, Schwarting RK. Dorsal hippocampal lesions boost performance in the rat sequential reaction time task. Hippocampus 2012;22:1202–14. Engel BT, Thorne PR, Quilter RE. On the relationships among sex, age, response mode, cardiac cycle phase, breathing cycle phase, and simple reaction time. J Gerontol 1972;27:456–60. Evans SM, Funderburk FR, Griffiths RR. Zolpidem and triazolam in humans: behavioral and subjective effects and abuse liability. J Pharmacol Exp Ther 1990;255:1246–55. Evenden JL, Turpin M, Oliver L, Jennings C. Caffeine and nicotine improve visual tracking by rats: a comparison with amphetamine, cocaine and apomorphine. Psychopharmacology 1993;110:169–76. Hartmann E, Orzack MH, Branconnier R. Sleep deprivation deficits and their reversal by d- and l-amphetamine. Psychopharmacology 1977;53:185–9. Hienz RD, Brady JV, Gooden VL, Vazquez ME, Weed MR. Neurobehavioral effects of head-only gamma radiation exposure in rats. Radiat Res 2008;170:292–8. Hienz RD, Weerts EM. Baboons in drug abuse research. In: VandeBerg JL, WilliamsBlanger S, Tardif SD, editors. The baboon in biomedical research. New York, NY: Springer; 2009. Jewett ME, Dijk DJ, Kronauer RE, Dinges DF. Dose–response relationship between sleep duration and human psychomotor vigilance and subjective alertness. Sleep 1999;22:171–9. Kenagy DN, Bird CT, Webber CM, Fischer JR. Dextroamphetamine use during B-2 combat missions. Aviat Space Environ Med 2004;75:381–6. Killgore WD, Rupp TL, Grugle NL, Reichardt RM, Lipizzi EL, Balkin TJ. Effects of dextroamphetamine, caffeine and modafinil on psychomotor vigilance test performance after 44 h of continuous wakefulness. J Sleep Res 2008;17:309–21. Kribbs NB, Pack AI, Kline LR, Getsy JE, Schuett JS, Henry JN, et al. Effects of one night without nasal CPAP treatment on sleep and sleepiness in patients with obstructive sleep apnea. Am Rev Respir Dis 1993;147:1162–8. Li XH, Wang JY, Gao G, Chang JY, Woodward DJ, Luo F. High-frequency stimulation of the subthalamic nucleus restores neural and behavioral functions during reaction time task in a rat model of Parkinson’s disease. J Neurosci Res 2010;88:1510–21. Lim J, Dinges D. Sleep deprivation and vigilant attention. Ann NY Acad Sci 2008a;1129:305–22. Lim J, Dinges DF. Sleep deprivation and vigilant attention. Ann NY Acad Sci 2008b;1129:305–22. Lim J, Wu WC, Wang J, Detre JA, Dinges DF, Rao H. Imaging brain fatigue from sustained mental workload: an ASL perfusion study of the time-on-task effect. NeuroImage 2010;49:3426–35. Loomis S, McCarthy A, Baxter C, Kellett DO, Edgar DM, Tricklebank M, et al. Distinct pro-vigilant profile induced in rats by the mGluR5 potentiator LSN2814617. Psychopharmacology (Berl) 2015;232:3977–89. Loomis S, McCarthy A, Edgar D, Tricklebank M, Gilmour G. Behavioural evidence that modafinil and amphetamine do not produce equivalent qualities of wake promotion in sleep-restricted rats. Sleep Med 2013;14(Supplement 1):e185. Macmillan NA, Creelman CD. Detection theory: a user’s guide. Cambridge: Cambridge University Press; 1991. Magill RA, Waters WF, Bray GA, Volaufova J, Smith SR, Lieberman HR, et al. Effects of tyrosine, phentermine, caffeine d-amphetamine, and placebo on cognitive and motor performance deficits during sleep deprivation. Nutr Neurosci 2003;6:237–46. May BJ, Huang AY, Aleszczyk CM, Hienz RD. Design and conduct of sensory experiments for domestic cats. In: Klump GM, Dooling RJ, Fay RR, editors.

71

Methods of comparative acoustics. Basil/Switzerland: Birkhauser Verlag; 1995. p. 95–108. Mayfield RD, Randall PK, Spirduso WW, Wilcox RE. Selective D1 and D2 dopamine receptor antagonists produce differential effects on reaction time in the rat. Pharmacol Biochem Behav 1993;46:759–68. Moody DB. Reaction time as an index of sensory function. In: W.C.S., editor. Animal psychophysics, the design and conduct of sensory experiments. New York, NY: Appleton-Century-Crofts; 1970. p. 277–302. Muir JL, Everitt BJ, Robbins TW. The cerebral cortex of the rat and visual attentional function: dissociable effects of mediofrontal, cingulate, anterior dorsolateral, and parietal cortex lesions on a five-choice serial reaction time task. Cereb Cortex 1996;6:470–81. Noble CE, BAKER EL, Jones TA. Age and sex parameters in psychomotor learning. Percept Mot Skills 1964;19:935–45. Oonk M, Davis CJ, Krueger JM, Wisor JP, Van Dongen HP. Sleep deprivation and timeon-task performance decrement in the rat psychomotor vigilance task. Sleep 2015;38:445–51. Paterson NE, Ricciardi J, Wetzler C, Hanania T. Sub-optimal performance in the 5-choice serial reaction time task in rats was sensitive to methylphenidate, atomoxetine and d-amphetamine, but unaffected by the COMT inhibitor tolcapone. Neurosci Res 2011;69:41–50. Pfingst BE, Hienz RD, Kimm J, Miller JM. Reaction-time procedure for measurement of hearing. I. Suprathreshold functions. J Acoust Soc Am 1975;57:421–30. Phillips JM, Brown VJ. Reaction time performance following unilateral striatal dopamine depletion and lesions of the subthalamic nucleus in the rat. Eur J Neurosci 1999;11:X1003–10. Pirch JH. Effects of dextroamphetamine on event-related slow potentials in rat cortex during a reaction time task. Neuropharmacology 1980;19:365–70. Rabbitt P. Psychological refractory delay and response-stimulus interval duration in serial, choice-response tasks. Acta Psychol 1969;30:195–219. Raymann RJ, Van Someren EJ. Time-on-task impairment of psychomotor vigilance is affected by mild skin warming and changes with aging and insomnia. Sleep 2007;30:96–103. Silber BY, Croft RJ, Papafotiou K, Stough C. The acute effects of d-amphetamine and methamphetamine on attention and psychomotor performance. Psychopharmacology 2006;187:154–69. Skinner BF. The behavior of organisms: an experimental analysis. New York, NY: Appleton-Century-Crofts; 1938. Skinner BF. Differential reinforcement with respect to time. Am Psychol 1946;1:274–5 (Abstr.). Smith NJ, Horst NK, Liu B, Caetano MS, Laubach M. Reversible Inactivation of rat premotor cortex impairs temporal preparation, but not inhibitory control, during simple reaction-time performance. Front Integr Neurosci 2010;4:124. Stebbins WC. Auditory reaction time and the derivation of equal loudness contours for the monkey. J Exp Anal Behav 1966;9:135–42. Stebbins WC. Response latency as a function of amount of reinforcement. J Exp Anal Behav 1962;5:305–7. Stebbins WC, Lanson RN. Response latency as a function of reinforcement schedule. J Exp Anal Behav 1962;5:299–304. Stebbins WC, Miller JM. Reaction time as a function of stimulus intensity for the monkey. J Exp Anal Behav 1964;7:309–12. Troy SM, Lucki I, Unruh MA, Cevallos WH, Leister CA, Martin PT, et al. Comparison of the effects of zaleplon, zolpidem, and triazolam on memory, learning, and psychomotor performance. J Clin Psychopharmacol 2000;20:328–37. Tucker AM, Basner RC, Stern Y, Rakitin BC. The variable response-stimulus interval effect and sleep deprivation: an unexplored aspect of psychomotor vigilance task performance. Sleep 2009;32:1393–5. Van Dongen HP, Dinges DF. Sleep, circadian rhythms, and psychomotor vigilance. Clin Sports Med 2005;24:237–49 (vii–viii). Van Dongen HP, Price NJ, Mullington JM, Szuba MP, Kapoor SC, Dinges DF. Caffeine eliminates psychomotor vigilance deficits from sleep inertia. Sleep 2001;24:813–9. Walker JL, Walker BM, Fuentes FM, Rector DM. Rat psychomotor vigilance task with fast response times using a conditioned lick behavior. Behav Brain Res 2011;216:229–37. Ward NM, Sharkey J, Marston HM, Brown VJ. Simple and choice reaction-time performance following occlusion of the anterior cerebral arteries in the rat. Exp Brain Res 1998;123:269–81. Warm JS, Parasuraman R, Matthews G. Vigilance requires hard mental work and is stressful. Hum Factors 2008;50:433–41. Waters WF, Magill RA, Bray GA, Volaufova J, Smith SR, Lieberman HR, et al. A comparison of tyrosine against placebo, phentermine, caffeine, and d-amphetamine during sleep deprivation. Nutr Neurosci 2003;6:221–35. Weed MR, Taffe MA, Polis I, Roberts AC, Robbins TW, Koob GF, et al. Performance norms for a rhesus monkey neuropsychological testing battery: acquisition and long-term performance. Brain Res Cogn Brain Res 1999;8:185–201. Welford A. Choice reaction time: basic concepts. Reaction Times 1980:73–128. Wesensten NJ, Killgore WD, Balkin TJ. Performance and alertness effects of caffeine, dextroamphetamine, and modafinil during sleep deprivation. J Sleep Res 2005;14:255–66. Whelan R. Effective analysis of reaction time data. Psychol Rec 2008;58:9. Wilkinson RT. Response-stimulus interval in choice serial reaction time: interaction with sleep deprivation, choice, and practice. Q J Exp Psychol A 1990;42:401–23.

A rodent model of the human psychomotor vigilance test: Performance comparisons.

The human Psychomotor Vigilance Test (PVT) is commonly utilized as an objective risk assessment tool to quantify fatigue and sustained attention in la...
565B Sizes 1 Downloads 10 Views