Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The ability of ethanol to impair working memory in humans (Maling, 1970; Roehrs et al., 1994) and to disrupt memory performance in laboratory animals is well known. The most frequently reported ethanol effect on memory in laboratory animals is a disruption in spatial working memory (Melchior et al., 1993; Givens and McMahon, 1995; Matthews et al., 1995). However, exactly which of the steps in the memory process (input, temporary storage, or recall) is disrupted by ethanol is not known. A disruption at any point in the process is observed as a decrement in the subject's working memory. Thus, if a drug disrupted attention it could disrupt the stimulus input and thereby affect memory processes.

Several studies have looked at the effects of ethanol on attention with various results. In humans, ethanol has been shown to disrupt (Leigh et al., 1977; Nelson and Wasserman, 1978; Jansen et al., 1985; Michel and Bättig, 1989; Koelega, 1995), have no effect (Talland, 1966; Dick et al., 1992; Roehrs et al., 1992; Lemon et al., 1993), or in one case improve attention (Doctor et al., 1966). Less research has been conducted on the effects of ethanol on attention in laboratory animals. Two previous studies reported that ethanol impaired the ability of rats to sustain attention (Givens, 1997; Givens and McMahon, 1997).

In the present study, dose-response curves for ethanol in pigeons responding under a model of working memory were compared with ethanol dose-response curves in pigeons responding under two models of attention. If the effects of ethanol on attention occur at doses equal to or lower than those that disrupt working memory, it would suggest that the disruption of working memory by ethanol is, at least in part, mediated by a disruption of stimulus input. Conversely, if the effects on attention were only observed at doses above those that disrupted working memory, the results would suggest that the effects of ethanol on working memory are not mediated by a failure of stimulus input and may be related to information storage or retrieval.

Working memory was studied by using a titrating matching-to-sample (TMTS) procedure (Wenger and Wright, 1990; Wenger et al., 1993). Under this schedule, a subject is presented with a stimulus (sample) that is then removed, and after a delay the animal must respond to the comparison stimulus that matches the sample stimulus. Thus, the procedure requires the recall of a stimulus that is not present at the time of the recall response. Accuracy is dependent upon the length of the delay, and accuracy in pigeons is reduced by drugs that also impair short-term memory in humans (Nelson and Wasserman, 1978; Spetch and Treit, 1986; Givens, 1997).

The effects of ethanol on attention were determined by using two different procedures. The first was a continuous-trial procedure in which the subject must maintain a continuous state of attention (with the exception of food presentation) for the entire test session. This procedure has many features similar to the 5-hole choice reaction time task (Muir et al., 1994; Jones et al., 1995). The second and more widely used attention task was a discrete-trial procedure similar to that used by Bushnell et al. (1997) and McGaughy and Sarter (1995). In this model, the subject must remain attentive for short-time periods throughout the test session. By using two different procedures, we hoped to obtain a better estimate of the effects of ethanol on attention and thus facilitate the comparison to the effects of ethanol on working memory. In addition, it would allow for a direct comparison of two procedures used to measure attention in laboratory animals. In some respects, the continuous-trial procedure may be a better model because the sustained level of attention required to perform the task is similar to that observed in the workplace. In this method, there is no beginning or end of a trial but rather the signal presentation is repeated throughout the test session without warning. However, the data generated by the continuous-trial procedure are much more difficult to analyze with signal-detection analysis because this method does not include the presentation of nonsignal trials. Therefore, the procedure lacks an accurate estimate of false alarm rates. The discrete-trial procedure is readily analyzed by signal-detection analysis, but it has a greater memory component and may not accurately reflect the workplace situation as well.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Subjects. Twenty-four adult male White Carneux pigeons (Palmetto Pigeon Plant, Sumter, SC) were housed in individual cages. The animals were maintained at 80 to 85% of their free-feeding body weight throughout all of the experiments by post-session feeding of Purina Checkers. Water was available ad libitum in the home cages. The animals were maintained on a 12-h light/dark cycle (lights on from 7:00 AM to 7:00 PM each day).

Apparatus. The test chamber was a Gerbrands Model G7313 pigeon test cage (Ralph Gerbrands Co., Arlington, MA). Mounted on the front panel of the chamber were three response keys that could be transilluminated with various colored lights. A response was defined as an opening of the key contact that occurred when the pigeon applied 0.15 N of pressure to the key in the form of a peck. A response operated a relay that produced an audible click (feedback). The test chambers were located in sound- and light-attenuating environmental enclosures (Ralph Gerbrands Co.). A motorized fan located within the environmental chamber provided ventilation and white noise. The test chamber was illuminated with a 28-V d.c. lightbulb. An IBM-compatible microcomputer installed with MedState software (MedAssociates, East Fairfield, VT) controlled the experiments and recorded the data.

Behavioral Training and Final Schedule: TMTS. Eight subjects were studied under a TMTS schedule. These animals had a previous drug history, including benzodiazepines, barbiturates, and amphetamine under this same schedule with no evidence of tolerance development or sensitization. The subjects were initially trained to eat Purina Checkers from a lighted feeder. Subsequently, the subjects were trained to respond upon the center key (transilluminated with the sample stimulus: either a red or green light) for food presentation by the process of successive approximations. A response on the center key resulted in 5-s access to food. Gradually, the number of pecks required on the center key to obtain access to food was increased to 15. Subsequently, responding on the center key under a fixed ratio 15 schedule resulted in a 3-s delay during which the lights transilluminating all three of the keys were extinguished. Following the delay period, two of the three keys, randomly selected, were illuminated: one with a red and one with a green light (comparison stimuli). A single response on the key illuminated with the same color as the sample stimulus was defined as a correct matching response and resulted in 5-s access to Purina Checkers. A response on the key illuminated with the color that was not presented during the presentation of the sample was an incorrect match and resulted in a 5-s time-out period during which all lights in the chamber were extinguished. A response on the dark key was counted but had no consequence. Immediately following access to food or the time-out, the center key was once again transilluminated with the red or green light.

Data Collection: TMTS. The data collected during each session included the following: mean delay value (sum of the delay values for each trial divided by the total number of trials), maximum delay value (maximum delay the subjects attained during a test session), total number of trials completed, percentage of correct matching responses (total number of correct matching responses divided by the total number of trials multiplied by 100), and response rates (number of responses on the center key in the presence of the sample stimulus divided by the total amount of time that the sample stimulus was presented). Any session in which a subject did not complete a minimum of 10 trials was eliminated from the group data analysis for mean delay, maximum delay, and percentage of correct matching response.

Behavioral Training and Final Schedule: Discrete Trial. Eight pigeons, with a prior exposure to three doses of ethanol during the determination of ethanol blood levels (see below), were trained to respond under a discrete-trial attention procedure. The subjects were first trained to eat Purina Checkers from an illuminated feeder. Next, the subjects were trained to respond for food presentation on each of the three keys by successive approximation. During the training session, one of the three keys was randomly illuminated with a red light, and the other two keys remained dark. (Each of the three keys was illuminated an equal number of times.) The key remained illuminated until the subject pecked the key at which time the animal received 5-s access to food. The training session terminated after the subject received 60 food presentations. After all the subjects were responding on all three keys, attention training was initiated. At the start of each trial, all three keys were transilluminated with green lights, and the house light was illuminated. Following an interval of time averaging 6.5 s (range 3-10 s), a signal was presented. The signal was defined as the stimulus color of one of the three keys changing from green to red. A postsignal interval (PSI) during which all three keys were illuminated with green lights followed the presentation of the signal. Initially, the duration of the signal was set at 10 s and the PSI at 0.1 s. Over the next 100 training sessions, the duration of the signal was gradually reduced to 250 ms and the PSI lengthened to 2 s. The signal duration of 250 ms was determined in pigeons responding under the continuous-trial procedure as the signal duration that produced a sensitivity index (SI) of 0.8. After the PSI, the green lights transilluminating the three response keys were extinguished, the house light was extinguished, and the two side keys were transilluminated (the right key with a white light and the left key with a blue light). For half of the pigeons, following signal trials, a response on the right key was defined as a hit, and a response on the left key was defined as a miss. Following blank trials (a trial during which no signal was presented), a response upon the left key was considered a correct rejection, and a response on the right key was considered a false alarm. For the other half of the pigeons, the key conditions were reversed. A hit or a correct rejection produced 3-s access to Pigeon Purina Checkers followed by the extinguishing of all lights in the chamber for an additional 7 s (the intertrial interval). At the end of the intertrial interval, the next trial began. A miss or a false alarm or the completion of a 3-s period following the PSI without a response (error of omission) produced a 10-s period during which all lights in the chamber were extinguished before the start of the next trial. When the subjects' behavior stabilized (~26 weeks), the effects of ethanol were determined. The test session terminated after the completion of 60 trials.

Data Collection: Discrete Trial. The data collected during each session included the total number of hits, false alarms, correct rejections, misses, errors of omission, signal trials, nonsignal trials, and the latency to respond following the completion of the PSI. Hits, false alarms, correct rejections, misses, and errors of omission were expressed as probability ratios (i.e., the probability of a hit was determined by taking the total number of hits for the test session and dividing by the total number of signal trials, and the probability of a correct rejection was determined by dividing the number of correct rejections by the total number of blank trials). The SI was calculated according to the method of Appel and Dykstra (1977) for discrete-trial procedures. Thus, SI for the discrete-trial procedure equals:

(1)

where h = probability of a hit and f = probability of a false alarm.

Behavioral Training and Final Schedule: Continuous Trial. Eight drug-naïve pigeons were trained to respond under a continuous-trial attention procedure. The subjects were first trained to eat Purina Checkers from an illuminated feeder. Next, the subjects were trained to respond for food presentation on each of the three keys by successive approximation. During the training session, one of the three keys was randomly illuminated with a red light, and the other two keys remained dark. (Each of the three keys was illuminated an equal number of times.) The key remained illuminated until the subject pecked the key at which time the animal received 5-s access to food. The training session terminated after the subject received 60 food presentations. After all the subjects were responding on all three keys, attention training was initiated. At the start of the test session, all three keys were transilluminated with green lights, and the house light was illuminated. Following an interval of time averaging 6.5 s (range of 3-10 s), a signal was presented. The signal was defined as the stimulus color of one of the keys changing from green to red. A response on the red key during the signal presentation or during the 2 s following the termination of signal was defined as a hit and resulted in food presentation. A response on one of the two green keys not associated with the signal during the signal presentation or the 2 s following the signal was defined as an incorrect response. The failure to respond during the signal or the 2 s following the signal was defined as a miss. A response on any green key in the absence of a signal was defined as a false alarm. Incorrect responses, misses, and false alarms had no scheduled consequences.

Signal Duration Determination. The subjects were trained to detect signal durations of 25, 50, 100, 250, 500, 750, and 1000 ms. The order of presentation for each signal duration was random for each subject. When stability (no significant trends over 10 successive daily sessions) had been achieved at a given signal duration, the mean value of the SI was determined and training proceeded with a different duration. These data were used to select the appropriate signal duration for the remainder of the experiments. Following the determination of the signal duration, all the pigeons were trained at the selected signal duration. When the subjects' behavior stabilized (~2 weeks) at this signal duration the effects of ethanol were determined. The test session terminated after 60 signal presentations.

Data Collection: Continuous Trial. The data collected during each session included the total number of hits, incorrect responses, misses, false alarms, and the overall response rate (sum of hits, false alarms, and incorrect responses divided by the session length minus the time for food presentation). Hits, misses, false alarms, and incorrect responses were expressed as probability ratios. SI was calculated by using the modification of signal detection analysis proposed by Sahgal (1988) for continuous-trial procedures. This modification assumes that the general lever pressing activity of the subject during the periods between signal presentation is a valid measure of false alarm rates. This method makes no assumptions about the distribution of the SI relative to the background noise and thus uses a nonparametric analog of the traditional sensitivity measure. Thus, SI for the continuous-trial procedure is equal to:

(2)

where h = probability of a hit, m = probability of a miss, and f = probability of a false alarm.

Ethanol Blood Levels. Eight drug-naïve pigeons (the same pigeons were later studied under the discrete-trial baseline) were given three doses of ethanol (0.3, 1, and 3 g/kg) by the intragastric route. Ethanol was administered as a 15% (w/v) solution, and each dose was separated by at least 5 days. A mixed order of dosing was used with each subject receiving a different dose on any given experimental day. Blood (10 µl) was collected from the wing vein at 45 and 75 min following administration of ethanol. Previous experiments had shown that blood ethanol levels peaked in pigeons at approximately 45 min after intragastric administration.

Ethanol Dosing. A 15% concentration (w/v) of ethanol was prepared from 100% ethanol and deionized water. Acute dose-response curves were determined by using five ethanol doses ranging from 0.3 to 3 g/kg. Ethanol was administered by the intragastric route twice a week (typically, Tuesdays and Fridays). The volume of the 15% ethanol solution was adjusted for each pigeon to achieve the desired dose. Vehicle (deionized water) was administered once a week (typically, Thursdays) via the intragastric route in a volume of 1 ml/kg. Because previous pharmacokinetic experiments determined that peak ethanol blood levels occurred ~45 min after dosing, ethanol was administered 45 min before the start of the test session. The animals were placed in the test cage during the 45 min before the start of the test session. A mixed order of dosing was used with each subject receiving a different dose on any given experimental day.

Statistical Analysis. In the TMTS procedure, a repeated measures ANOVA was used to determine the presence of significant drug effects on percentage of correct matching response, mean and maximum delay values, and rate of responding. In the discrete-trial attention procedure, a one-way repeated-measures ANOVA was used to determine significant drug effects on the probabilities of a hit, miss, false alarm, correct rejection, error of omission, as well as latency to respond following the PSI. Similarly, in the continuous-trial procedure, a one-way repeated-measures ANOVA was used to determine significant drug effects on the probabilities of a hit, miss, false alarm, incorrect response, as well as response rates. Under all the experiments, if data failed the test for equal variance, a Friedmann's repeated-measures analysis on ranks was used. All data were subjected to a post hoc Dunnett's test. A P value of <.05 was required for significance. A two-way ANOVA followed by a post hoc Student-Newman-Keuls test was used to determine significance of ethanol levels in blood samples. A P value of <.05 was required for significance.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Blood Ethanol Levels. Blood ethanol concentrations (Table 1) were determined at two time points following intragastric administration of three different doses of 15% (w/v) ethanol. At each time point, there was a significant dose-dependent increase of ethanol in the blood (P < .05). However, no significant difference was observed between blood levels at the 45- and 75-min time points. Thus, there was no significant change in the blood ethanol level during the first 30 min of the session. This is significant because most of the experimental test sessions were completed within 30 min. The only exceptions to this occurred at 1.8 and 3 g/kg of ethanol under the TMTS procedure at which four of eight subjects and seven of eight subjects, respectively, required >30 min to complete the test session.

TMTS. Control performance and the effects of ethanol on TMTS performance are shown in Fig. 1. Control values revealed that subjects maintained ~80% matching accuracy as predicted by the titration procedure. The average maximum delay was 26.2 ± 2.3 s and the average mean delay value was 10.9 ± 1.1 s. The control rate of responding to the sample stimulus was 1.5 ± 0.2 responses/s.

View larger version (13K): [in this window] [in a new window]   Fig. 1.   Effect of ethanol (15% w/v) on the performance of pigeons responding under the TMTS procedure. Abscissa, dose of ethanol on log scale; ordinate (top), mean and maximum delay values for the session expressed in seconds; ordinate (middle), total percentage of matching accuracy for the session; ordinate (bottom), rate of responding to the sample stimulus expressed in responses per second. Data points and error bars above V represent the vehicle injection ± S.E. Data points and error bars for the effects of ethanol represent the mean ± S.E. of individual determinations in eight pigeons unless otherwise indicated by (n). *P < .05 compared with vehicle.

Attention: Signal Duration. Eight pigeons were trained to respond under the continuous-trial schedule of signal presentation. The pigeons were presented signals with durations ranging from 25 to 1000 ms. As can be seen in Fig. 2 (top), the probability of a hit [p(hit)] increased as the signal duration increased up to a duration of 100 ms. No additional increases were observed at longer signal durations. As with the p(hit), with increasing signal duration, the probability of an incorrect response [p(incorrect response)], probability of a false alarm [p(false alarm)], and probability of a miss [p(miss)] decreased. The SI was calculated for the data (Fig. 2, bottom). The SI value increased up to a duration of 100 ms. The 250-ms signal duration was selected as the signal duration for the remainder of the continuous-trial experiments. This duration also was selected for the discrete-trial experiments to facilitate the comparison between performance under the continuous and discrete-trial procedures. This signal duration produced a SI that was not significantly different from the 100-ms signal duration but had less variability about the mean.

View larger version (18K): [in this window] [in a new window]   Fig. 2.   Effect of increasing signal duration on the performance of pigeons responding under the continuous-trial attention procedure. Abscissa, signal duration expressed in milliseconds; ordinate (top), probability values (number of responses divided by the total number of signals); ordinate (bottom), sensitivity index. Data points and error bars represent the mean ± S.E. of individual determinations in five pigeons. *P < .05.

Discrete Trial. Control performance and the effects of ethanol on performance under the discrete-trial attention task are shown in Fig. 3. Under control conditions, the p(hit) (0.81 ± 0.05) and probability of a correct rejection [p(correct rejection)] (0.87 ± 0.50) were high, whereas the p(false alarm) (0.11 ± 0.04), p(miss) (0.16 ± 0.06), and probability of an error of omission [p(error of omission)] (0.03 ± 0.02) were low. Latency to respond was 0.69 ± 0.08 s. Only the highest dose of ethanol (3 g/kg) caused a significant decrease in the p(hit) and p(correct rejection) as well as a significant increase in p(error of omission). However, ethanol did not have any significant effects on the p(false alarm) or p(miss). At the 3-g/kg dose, the latency to respond and the number of errors of omission dramatically increased. Thus, attention as measured by the p(hit) was only impaired at a dose that significantly affected the ability of the animal to respond. However when the sensitivity index was calculated for the data, it was found that the two highest doses (1.8 and 3 g/kg) of ethanol impaired the ability of the animals to distinguish the signal from background noise.

View larger version (14K): [in this window] [in a new window]   Fig. 3.   Effect of ethanol (15% w/v) on the performance of pigeons responding under the discrete-trial attention procedure. Abscissa, dose of ethanol on log scale; ordinate (top), probability values (number of responses divided by the total number of signal or nonsignal trials); ordinate (middle), sensitivity index; ordinate (bottom), latency to respond expressed in seconds. Data points and error bars above V represent the vehicle injection ± S.E. Data points and error bars for the effects of ethanol represent the mean ± S.E. of individual determinations in eight pigeons unless otherwise indicated by (n). *P < .05 compared with vehicle.

Continuous Trial. Under the continuous-trial procedure, p(hit) (0.96 ± 0.01) was nearly perfect under control conditions (Fig. 4). The p(miss) and p(incorrect response) were nearly zero, whereas the p(false alarm) (0.22 ± 0.09) was slightly higher. Even though the continuous-trial SI was calculated by using a nonparametric analog of the SI equation described for the discrete-trial procedure, the control values for SI under both procedures were similar (discrete trial, SI = 0.73 ± 0.06; continuous trial, SI = 0.79 ± 0.07). Similar to the discrete-trial procedure, a dose-dependent effect was observed when ethanol was administered to subjects trained under the continuous-trial procedure. Ethanol caused a significant decrease in the p(hit) at the two highest doses. A large increase in p(false alarms) was observed that peaked after 1.8 g/kg. There was also a significant increase in the p(miss) at all doses compared with control conditions, but no significant effect of ethanol was observed on the p(incorrect response) made by the subjects (Fig. 4, top). The increase in false alarms made by the subjects corresponded to a slight increase in the animals' response rate (Fig. 4, bottom), but the increase in response rate was not statistically significant. The SI value was calculated for the data, and it was found that ethanol did not affect the sensitivity of the subjects to the signal (Fig. 4, middle). This was an unexpected result especially because ethanol had such a pronounced effect on hits, misses, and false alarms, and because it produced significant effects on sensitivity under the discrete-trial procedure.

View larger version (14K): [in this window] [in a new window]   Fig. 4.   Effect of ethanol (15% w/v) on the performance of pigeons responding under the continuous-trial attention procedure. Abscissa, dose of ethanol on log scale; ordinate (top), probability values (number of responses divided by the total number of signals); ordinate (middle), sensitivity index; ordinate (bottom), response rate expressed as responses per minute. Data points and error bars above V represent the vehicle injection ± S.E. Data points and error bars for the effects of ethanol represent the mean ± S.E. of individual determinations in eight pigeons unless otherwise indicated by (n). *P < .05 compared with vehicle.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The impairment of working memory by ethanol was demonstrated by dose-dependent decreases in accuracy and both maximum delay and mean delay values for the session as measured by the TMTS procedure. Delay values were decreased at a dose (1 g/kg) that did not reduce overall matching accuracy <80%. The TMTS program compensated for a decreased accuracy at long delay values by lowering the delay to a value at which the animal could still perform at 80% accuracy. Delay values were even shorter at the 1.8- and 3-g/kg doses.

Other laboratories have shown similar results by using various animal models of working or short-term memory (Castellano and Pavone, 1988; Melchior et al., 1993; Givens, 1995; Matthews et al., 1995; Givens and McMahon, 1997). Several groups have shown that ethanol causes a dose-dependent (Melchior et al., 1993; Givens, 1995; Matthews et al., 1995; Givens and McMahon, 1997) and delay-dependent (Givens, 1995; Givens and McMahon, 1997) decrease in working memory as tested in animals performing in t-mazes, radial arm mazes, choice reaction time tasks, and delayed matching-to-sample procedures. Passive avoidance procedures also have shown a decrease in memory retention following the administration of ethanol (Holloway, 1972; Bammer and Chesher, 1982; Castellano and Pavone, 1988; Sasaki et al., 1995).

There are significant procedural differences between the two measures of attention, and it is interesting to compare control performance and the effects of ethanol under the two procedures. Under control conditions, when the signal duration is held constant across procedures there is a significant difference in p(hit) and p(miss). However, although calculated differently, there is no difference in the measure of SI under the two procedures. There are also a number of significant differences in the effects of ethanol under the two procedures. Under the discrete-trial procedure only the highest dose (3g/kg) decreased p(hit). This effect was observed only at a dose that caused a disruption in the ability of the animal to perform as shown by the increase in the subjects' latencies to respond and an increase in errors of omission (lack of response on a signal or nonsignal trial). Under the continuous-trial procedure, the effects of ethanol on p(hit), p(miss), and p(false alarm) were observed at lower doses than those observed under the discrete-trial procedure (1.8 and 3 g/kg). In addition, ethanol produced an increase in false alarms not observed under the discrete-trial procedure. Finally, under the discrete-trial procedure ethanol had significant effects on response rate and on the ability of subjects to detect the signal as measured by the SI. These two effects were not observed under the continuous-trial procedure.

It is interesting to note that false alarms increased following ethanol under the continuous-trial procedure but not the discrete-trial procedure. Under the discrete-trial procedure, a false alarm is defined as the subject responding on the response key, indicating that the animal has detected the presence of a signal when in fact no signal has been presented. Under the continuous-trial procedure, a false alarm is defined as any response that occurred in the absence of the signal. However, under control conditions p(false alarm) values were not significantly different under the two procedures.

If the assumption that lever-pressing activity is a valid measure of false alarms under the continuous-trial procedure, it can be assumed that ethanol was disrupting the ability of the animal to detect the signal as indicated by a decrease in hits and an increase in false alarms. However, it could be the case that rather than ethanol inhibiting the ability of the animal to remain vigilant, it may be that ethanol is increasing response rate. At the 1.8-g/kg dose of ethanol, there is an increase in false alarms, which is associated with a slight (but not significant) increase in response rate. It is well known that as blood ethanol levels increase in the blood stream, ethanol can have a stimulant effect (Earleywine and Martin, 1993; Holdstock and de Wit, 1998). Thus, a possible explanation for the increase in false alarm rate is an ethanol-induced increase in responding. However, this possibility is complicated by the fact that 1.8 g/kg of ethanol did not significantly decrease response latency under the discrete-trial procedure. In addition, the increase in false alarms was observed at a dose that did not correspond to significant increases in response rate under the TMTS procedure. This increase in false alarms also occurred at a dose that decreased p(hit) and increased p(miss). Therefore, this effect may not be fully explained by a nonspecific increase in response rate.

Another difference between the two attention procedures is the effect of ethanol on the SI. Ethanol caused a significant decrease in the SI under the discrete-trial procedure, but there was no effect on sensitivity under the continuous-trial procedure. This may be an artifact of the SI equation for the continuous-trial procedure. However, it should be noted that the control values for SI were similar.

Overall, both procedures used to measure attention showed that the ability of the subjects to maintain a sustained level of attention was disrupted by ethanol. However, the way in which attention was disrupted by ethanol appeared to differ depending upon the procedure studied. Future drug studies directly comparing the two procedures may help in the interpretation of effects and validation of both procedures.

In the present study, a comparison of the dose-response curves for the effects of ethanol under the TMTS procedure and the two attention procedures reveals that doses that decreased matching accuracy on the memory task (1 and 1.8 g/kg) did not effect attention performance. Thus, it appears that the ability of ethanol to inhibit memory function is not the result of the subject's inability to remain alert to stimuli in the environment, and thus, the effect of ethanol on stimulus input is probably not a major factor in the effect of ethanol on working memory.

This result is in contrast to the results reported by Givens and McMahon (1997). In their study, ethanol was shown to disrupt both working memory and attention in the rat. They used a modified matching-to-sample procedure to simultaneously measure working memory and attention in the rat. Working memory in the Givens and McMahon (1997) procedure was measured as a change in choice key accuracy as a result of an increase in the delay between the sample and choice phases. Attention was measured as a change in matching accuracy as a function of a change in the length of the sample stimulus presentation (500 and 1000 ms) and also as a decrease in choice performance over time.

We feel that the simultaneous measurement of working memory and attention in the rat is not as easy as suggested in Givens and McMahon (1997). It is well known that sample stimulus duration is an important variable in delayed matching-to-sample performance (Riopelle, 1959; Nelson and Wasserman, 1978). However, it is unclear whether changes in signal duration affect attention or the strength of the memory trace.

To reach the conclusion drawn by Givens and McMahon (1997), it is important to first show that with a validated model of sustained attention, the performance of rats is significantly different when signal durations of 500 and 1000 ms are presented. The majority of the literature on attention effects in rats does not use signal durations >500 ms. Indeed those studies that do examine signal duration and include values as long as 500 ms show that performance is approaching nearly maximally attainable values at 500 ms. Therefore, it is difficult to believe that there would be a significant difference in attention performance at signal durations >500 ms. In the absence of comparative data in rats showing a significant difference in performance with a model of sustained attention with signal durations of 500 or 1000 ms, the Givens and McMahon (1997) interpretation is difficult to support.

It also should be noted that although the work was done in pigeons, the present study shows that under control conditions, there is no difference in attention at signal durations ranging from 250 to 1000 ms (Fig. 2). Attention is only changing significantly as a function of signal duration at values <250 ms.

Memory is a complex process as the conflicting results of these two studies demonstrates. Future research is necessary to help understand the processes involved in memory and the steps in the memory process that are specifically disrupted by ethanol.