- Email: [email protected]
The effect of cognitive load on saccadic eye movements
Acta Psychologica 104 (2000) 69±85
www.elsevier.com/locate/actpsy
The eect of cognitive load on saccadic eye movements Els Stuyven a
a,*
, Koen Van der Goten a, Andre Vandierendonck a, Kristl Claeys b, Luc Crevits b
Department of Experimental Psychology, University of Gent, Henri Dunantlaan 2, B-9000 Gent, Belgium b Department of Neurology, University of Gent, Belgium Received 22 February 1999; received in revised form 20 September 1999; accepted 20 September 1999
Abstract The present study tested the hypothesis that, unlike prosaccades, antisaccades require controlled processing, due to the prepotent response that needs to be inhibited. The eect of the Random time Interval Generation (RIG) task (Vandierendonck, A., De Vooght, G., & Van der Goten, K. (1998). European Journal of Cognitive Psychology, 10, 413±444) on these saccade latencies and errors was studied. This task has the advantage that it loads executive processes, with only minimal interference with verbal or visuo-spatial components. A ®rst experiment compared saccade performance within the prosaccade and the antisaccade task, executed alone and in combination with the RIG task and ®xed tapping (added to exclude possible motor component interference explanations). A second experiment investigated the in¯uence of task characteristics on the eects found. Although it was shown that antisaccades are more prone to interference of an executive interference task, it seems that prosaccades are also vulnerable. Interference on prosaccades could originate from a controlled execution of these saccades. A third experiment con®rmed that endogenously generated prosaccades are susceptible to dual-task interference and showed that controlled saccade execution, without the need to inhibitit a prepotent response, is sucient to produce interference. Ó 2000 Elsevier Science B.V. All rights reserved. PsycINFO classi®cation: 2323; 2340; 2346
*
Corresponding author. Tel.: +32-(0)9-2646406; fax: +32-(0)9-2646496. E-mail address: [email protected] (E. Stuyven).
0001-6918/00/$ - see front matter Ó 2000 Elsevier Science B.V. All rights reserved. PII: S 0 0 0 1 - 6 9 1 8 ( 9 9 ) 0 0 0 5 4 - 2
70
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
1. Introduction Recent evidence suggests that eye movements towards a laterally presented stimulus (prosaccades) are performed automatically, whereas eye movements in a direction opposed to the normal saccadic direction (antisaccades) are performed under attentional control (Roberts, Hager & Heron, 1994). The antisaccade task was developed by Hallet (1978; Hallet & Adams, 1980) to investigate the mechanisms that underly the generation of automatic and voluntary eye movements. He found that antisaccadic eye movements typically have longer latencies (time between stimulus onset and the start of the eye movement), that inaccurate re¯exive saccades towards the cue sometimes occur, and that practice does not result in an improvement of antisaccade performance. 1.1. Antisaccades and frontal lobes Guitton, Buchtel and Douglas (1985) used the antisaccade task to investigate de®cient inhibitory control in patients with prefrontal lesions. The Frontal Eye Field (FEF, area 8) and the Colliculus Superior (CS) are two systems involved in the generation of saccadic eye movements. The CS is supposed to mediate automatic, re¯exive movements, whereas the FEF is assumed to be involved in higher level control of saccadic eye movements. The results of the frontal patients were compared with those of a control group and a group with lesions in the temporal lobe. No dierences between the groups were found in latencies or errors of prosaccades. In the antisaccade task, however, frontal patients made more re¯exive saccades (i.e., incorrect saccades towards the stimulus). Moreover, some frontal patients only performed a re¯exive saccade with stimulus presentation contralateral to the damaged hemisphere. The results suggest that the damaged structures are involved in the generation of the antisaccades. This strengthens the idea that the prefrontal cortex contributes to the generation of complex saccadic eye movement behaviour. The function of these lobes would then be the inhibition of re¯exive oculomotor activity and the triggering of voluntary eye movements. Other authors used the antisaccade task in populations where prefrontal dysfunctioning is hypothesised (e.g., schizophrenia, ADHD). Fukushima et al. (1988) studied prosaccade and antisaccade performance in schizophrenic patients. Compared to healthy control participants, no dierences were observed in the prosaccade condition. However, half of the schizophrenic patients made more incorrect antisaccades, and the correct antisaccades had longer latencies. Fukushima, Fukushima, Morita and Yamashita (1990) presented schizophrenic patients with a nosaccade task and a memory saccade task. In the nosaccade task, participants ®xate a central point in spite of the presentation of a cue. In the memory saccade task, a cue is presented while the ®xation point is still present. Only after the removal of the ®xation point a saccade must be made to the point where the cue was previously presented. Schizophrenic patients made a high number of saccades in the nosaccade task, whereas the healthy controls made no mistakes. Moreover, the memory saccade latencies were longer in schizophrenic patients than in the controls.
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
71
In view of this evidence it seems that schizophrenic patients, much like prefrontal patients, have diculty inhibiting re¯exive saccades and executing voluntary saccades when the target position is known but not visible. The connection with the frontal lobes is also supported by the work of Rosse, Schwartz, Kim and Deutsch (1993) who found a signi®cant correlation between performance on the antisaccade task and the Wisconsin Card Sorting Task in schizophrenic patients. Crevits and De Ridder (1997) studied antisaccades and memory saccades in a large group of Parkinson's patients. They demonstrated severe abnormalities suggesting a dysfunction in the striato-prefrontal lobe. However, previous reports in smaller groups of less severely aected patients could not show consistent abnormalities (Kitagawa, Fukushima & Tashiro, 1994; Fukushima, Fukushima, Miysaka & Yamashita, 1994). 1.2. Antisaccades and working memory The multi-component working memory model proposed by Baddeley and Hitch (1974) and Baddeley (1986, 1990), has proven to be a very fruitful conceptualisation of short-term memory. It has earned its merits in various areas of cognitive psychology such as reading comprehension, mental arithmetic, imagery, reading development, problem solving and others. Moreover, the model has demonstrated its usefulness in the study of patients with neurological de®cits. It has, for instance, been found that interference with attentional control in dual-tasks, results in patterns that are similar to those found in patients with prefrontal lesions (Baddeley, 1990). Roberts et al. (1994) investigated the relationship between antisaccades and working memory. Like Gathercole (1994), these researchers assume two important functions of the prefrontal cortex: (1) the short-term retention of information in order to organise future actions, and (2) the inhibition of prepotent responses. In addition, the authors assume that a temporary increase in working memory load can result in patterns of errors similar to patterns of permanent dysfunction of working memory, like those observed in prefrontal patients. Roberts et al. (1994) compared the single-task and dual-task performance in a prosaccade and an antisaccade task. The secondary tasks used were mental arithmetic, a shadowing task (i.e. repeat a given number), or a repeat task (i.e. always repeat the same number). Logie, Gilhooly and Wynn (1994) have already shown that mental arithmetic loads the central executive and the phonological loop. In this experiment Roberts et al. (1994), used only healthy participants. In the prosaccade task, almost no errors were made and mental arithmetic did not aect saccade latency. In the antisaccade task, however, more directional errors (re¯exive saccades towards the stimulus) occurred and saccade latencies increased, much like in the prefrontal patients of Guitton et al. (1985). These results suggest at least a behavioural similarity between frontal patients and non-patients under heavy working memory load. Roberts et al. (1994) assumed that as long as working memory is able to intervene in response preparation, alternative responses can be inhibited. The CS, which takes part in re¯exive saccades receives input directly (FEF±CS) and indirectly (FEFcaudate nucleus-substantia nigra-CS) from the dorsolateral prefrontal cortex. The working memory system can inhibit the CS, for instance to prevent re¯exive saccades
72
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
in the antisaccade task and it may also be involved in initiating a planned saccade. However, the more working memory resources are involved in other activities, the less resources are available for inhibitory control and planning, and the more probable it will be for directional errors to occur in the antisaccade task. In order to show unequivocally that the central executive, rather than the entire working memory system, takes part in antisaccade preparation, the concurrent task should be speci®c: it should interfere with the central executive without loading the slave systems. The literature contains a large number of tasks that interfere with the central executive: e.g., random number generation (Baddeley, 1966), random digit generation (e.g., Gilhooly, Logie, Wetherick & Wynn, 1993), tapping a random key pattern (Rapee, 1993), avoidance of automaticity in a verbal reaction task (Greenwald & Shulman, 1973), random pursuit tracking (Dalrymple-Alford, Kalders, Jones & Watson, 1994), free recall (Rusted, 1988), counting backwards (e.g., Glanzer, Dorfman & Kaplan, 1981), and generation of category members (e.g., Baddeley, Lewis, Eldridge & Thomson, 1984). All these tasks have the drawback of also interfering with either the visuo-spatial or the phonological slave subsystem, so that clear-cut conclusions are not easy to obtain. Vandierendonck, De Vooght and Van der Goten (1998) have developed and tested a central executive task that is allegedly pure in the sense that it does not load signi®cantly either of the known slave systems. In this Random time Interval Generation (RIG) task, participants are to tap a randomly spaced sequence of time intervals on a key so as to produce a completely unpredictable ``rhythm''. In order to tap randomly, automaticity has to be prevented, and so this task requires cognitive control, while there are no logical or empirical reasons to assume that it interferes with the slave systems. This task has already been shown to aect span, supraspan (Vandierendonck et al., 1998), stimulus independent thoughts (Stuyven & Van der Goten, 1995), a dual content location span task (Martein, Kemps & Vandierendonck, 1999) and simple mental arithmetic (De Rammelaere, Stuyven & Vandierendonck, 1999). The reported experiments were inspired by the working memory model. However, because the concept of the ``central executive'' and its until recently presumed unitary nature stands to discussion, we prefer to use the more neutral term ``executive functions''. 2. Experiment 1 The eect of the RIG task on saccades was ®rst studied in Claeys et al. (1999). Prosaccades and antisaccades were executed in a single- and a dual-task (i.e. in combination with RIG) situation. On the basis of the hypothesis that prosaccades are automatic, the authors expected no interference of the secondary task with prosaccade performance. However, it turned out that both prosaccade and antisaccade latencies are in¯uenced by the RIG load and that there is no kind of interaction of saccade task and load. The question can be raised whether a possible eect of RIG on the saccades could be due to a central cognitive component, a motor component, or simply to the fact that two tasks have to be performed at the same time. A central component would involve the cognitive control over the actions to be
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
73
performed (i.e., the inhibition of a re¯exive saccade, the start of a saccade in the opposite direction of the stimulus, the creation of a random series of time intervals). A possible motor interference would re¯ect the diculty of performing two motorbased actions at the same time (i.e., eye movement and ®nger movement). In order to address these possibilities, in Experiment 1 a ®xed tapping task was included as an additional control condition for RIG as it requires the same motor actions, but assumably does not call on executive functions. In this case the output does not have to be monitored for randomness, and inhibition of familiar patterns is not needed. 2.1. Method 2.1.1. Participants and design Sixteen ®rst-year psychology students (seven female, nine male) of the University of Gent participated for course requirements and credit. They volunteered for this particular experiment. All participants had normal vision. A 2 (Saccade task: prosaccade or antisaccade) ´ 2 (2 Order of presentation) ´ 3 (Load: control, ®xed tapping, RIG) mixed factorial design was used in which the saccade task was a between-subjects and load a within-subjects variable. Participants were randomly assigned to the saccade groups and within this group to either of two randomly chosen order conditions (Fixed-Control-RIG and Control-RIG-Fixed). 2.1.2. Materials and procedure Eye movements were registered by means of an infra-red based eye tracking system, the Ober2 system. Two IBM-compatible PC's connected by means of a parallel cable were used in this experiment in order to synchronise eye movement registration and the secondary task. The program to present the visual stimuli and register the eye movements was written in the Nystr om language, which is part of the eye tracking software, and was used in every condition. Participants were seated at a distance of 600 mm from a 14 inch. screen. The sampling frequency was 300 Hz, so that a measurement of eye position was taken every 3.33 ms. Only the movements of the dominant eye were registered. In the centre of the screen a white square (4 4 mm or visual angle 0.38°) was used as the ®xation point. As in the experiments of Guitton et al. (1985), after 1.5±3 s, the square disappeared. At the same time another square of the same size appeared 70 mm to the left or to the right (visual angle 6.7°). After 3±5 s, the screen was cleared and the central ®xation square reappeared. The secondary RIG task was monitored by the second PC. Participants were instructed to tap an unpredictable sequence on the zero key of the numeric keypad of the computer at an average rate of one keystroke per second. It was emphasised that repetition of a pattern was not allowed. These tap sequences were registered from the moment eye movement registration was started until the program for eye movement registration sent a stop signal (one condition lasted about 3 min). For the ®xed tapping task, participants were instructed to hit the zero key of the numeric keyboard at a ®xed rate of one per second. As in the RIG task, tapping behaviour was registered.
74
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
Performance was measured in terms of latencies and errors of saccades. Reaction time is de®ned as the time interval between the appearance of the stimulus and the onset of the eye movement. Possible errors are: (1) saccades in the wrong direction; (2) unperformed saccades; (3) re¯exive saccades in the antisaccade task. Anticipatory saccades (latencies less than 100 ms or started before stimulus onset) were not counted as errors but were removed from the analysis (about 1% of the trials). After the instructions were given, the participant put on the infrared goggles of the Ober2 system and the room was darkened. At this moment, a calibration phase was started in order to adjust a number of parameters of the eye tracking apparatus (e.g., illuminance). Participants passed two calibration sessions of eight trials in which the prosaccade or antisaccade task was practised. There was also a practice session for both secondary tasks. After calibration, the three conditions were run. Every condition consisted of 30 trials, so that including calibration, every participant performed 106 saccades. 2.2. Results The data were analysed by means of a multivariate analysis. The eects of the between-subjects and the within-subjects variables and their interactions were estimated by means of contrasts between the dependent variables: the mean latencies per participant under each of the three load conditions. For the accuracy analysis the average number of errors per condition was the dependent variable. This procedure conforms to the suggestions of McCall and Appelbaum (1973) for a correct dataanalysis of repeated measures designs. 1 2.2.1. Eect of order The order of the conditions had no eect on the latencies or on the number of errors. The interactions with saccade task and load were not signi®cant either (all Fs < 1). 2.2.2. Latencies Fig. 1 displays the mean latencies for the prosaccades and the antisaccades under control, ®xed tapping, and RIG conditions. The main eect of saccade task was signi®cant with prosaccades performed faster (M 228 ms) than antisaccades (M 325 ms) (F(1, 10) 9.5, P < 0.05). The main eect of load was also signi®cant (Mcontrol 246, Mfixed 288, MRIG 295) (F(2, 9) 61.2, P < 0.001). The control condition diered from both the RIG and the ®xed tapping condition (F(1, 10) 41.8, P < 0.01 and F(1, 10) 14.9, P < 0.01, respectively). The dierence between the RIG and the ®xed condition, however, was not signi®cant F < 1. Furthermore, as predicted, the saccade task ´ load interaction was signi®cant 1
The data of the random and ®xed tapping tasks were analysed according to the method described by Vandierendonck et al. (1998). None of the experiments reported trade-os with the performance on the saccade tasks, hence, these analyses will not be discussed further.
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
75
Fig. 1. Latencies in ms for prosaccades (PS) and antisaccades (AS) as a function of secondary task conditions in Experiment 1.
(F(2, 9) 27.4, P < 0.001). Decomposition of this interaction revealed that the RIG versus control contrast interacted with the saccade tasks (F(1, 10) 6.2, P < 0.05). Fig. 1 clearly shows that the dierence between the control and the RIG condition was smaller (but signi®cant) in the prosaccade than in the antisaccade task. It is clear from this ®gure that the interaction of saccade task and the ®xed tapping versus control contrast was also signi®cant (F(1, 10) 19.5, P < 0.01). But contrary to our expectation, the interaction of saccade task with the random versus ®xed tapping contrast was not found F < 1. This pattern of results indicates that the ®xed tapping and the RIG task had comparable eects on the eye movement latencies. We will return to this in the discussion. Despite the interaction of saccade task and load, showing that antisaccades were more profoundly aected by the interference than prosaccades, a closer inspection revealed that in both the prosaccade and the antisaccade task, the eect of load was signi®cant (F(1, 5) 7.7, P < 0.05 and F(1, 5) 64.9, P < 0.001, respectively). 2.2.3. Errors Table 1 shows the average number of errors in the prosaccade and the antisaccade task under the dierent secondary task conditions. More errors occurred in antisaccades (M 5.29) than in prosaccades (M 0.08). F(1, 10) 15.9, P < 0.01. The Table 1
Mean number of errors per conditiona in Experiment 1
Prosaccade Antisaccade a
Control
Fixed tapping
RIG
0.00 (0.00) 3.88 (3.95)
0.25 (0.43) 4.50 (4.39)
0.00 (0.00) 7.50 (6.69)
Standard deviations are given in parentheses (RIG: random time interval generation task).
76
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
eect of load was also signi®cant (Mcontrol 1.31, Mfixed 2.38, MRIG 3.75) (F(1, 10) 7.4, P < 0.05), as was the task ´ load interaction (F(2, 9) 4.4, P < 0.05). This interaction was in line with the hypothesis: the dierence between RIG and control was larger in the antisaccade than in the prosaccade task (F(1, 10) 9.1, P < 0.05), as was the dierence between RIG and ®xed tapping (F(1, 10) 7.6, P < 0.05). Moreover, the interaction of saccade task with the contrast between ®xed tapping and control was not reliable (F(1, 10) 1.7, P > 0.20). 2.3. Discussion The results partially con®rm our hypothesis; antisaccade performance was aected more by the secondary task load than latencies and errors in the prosaccade task. However, two unexpected ®ndings arose from this experiment. First, although the interaction of saccade task and load was signi®cant, prosaccade latencies increased signi®cantly in the dual-task conditions. This is contrary to the hypothesis that prosaccades would be invulnerable, due to their automatic nature. One possible explanation is that participants actually executed their prosaccades in a controlled way in order to improve their performance. It could be that a controlled saccade execution makes these saccades vulnerable to interference. To support this idea, it is necessary to collect empirical evidence that the controlled execution of a saccade suces to produce interference with an executive task. Roberts et al. (1994) hypothesised that executive functions play a role in the planning of a saccade. However, by using the antisaccade task, they could only con®rm the interaction of the inhibition of a prepotent saccade and executive functions. Experiment 3 was designed to clarify this point. A second remarkable result is that there was no signi®cant dierence in the saccade latencies of the RIG and the ®xed tapping condition. It is possible that we observed motor interference rather than cognitive interference. In fact, the ®xed tapping task was included to control for motor interference, as the same motor actions are required without taxing executive functions. A cognitive interpretation of the eect remains possible, however. First, why would this result in a larger impairment of antisaccade performance if the interference is motor based? Second, the instructions given in the ®xed tapping condition were rather strict. Participants were told to hit the zero-key of the numeric keyboard every second. If participants took this task very seriously and monitored the time elapsed since the last key press, ®xed tapping might not be considered an automatic task, but a task which requires executive control for its accurate execution. 2 To test this hypothesis, Experiment 2 compared performance under strict instructions and more lenient instructions for the ®xed tapping task on both prosaccades and antisaccades under single- and dual-task conditions.
2
We are indebted to Gino De Vooght for this suggestion.
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
77
3. Experiment 2 3.1. Method 3.1.1. Participants and design Thirty two ®rst-year psychology students (26 female, 6 male) of the University of Gent participated for course requirements and credit. They volunteered for this particular experiment. All participants had normal vision, none of them had participated in Experiment 1. A 2 (Saccade task: prosaccade or antisaccade) ´ 2 (Load: control, ®xed tapping) ´ 2 (Instruction: strict, lenient) mixed factorial design was used with saccade task and instructions as between-subjects variables and with load as a within-subjects variable. Participants were randomly assigned to one of the four between-subjects cells. As Experiment 1 has shown that the condition order had no eect, only one order was used here. 3.1.2. Materials and procedure The materials were identical to those of Experiment 1. Every participant ®rst performed the control condition of the experiment. Both the control and the ®xed tapping condition were preceded by a practice session of eight saccades. The experimental sessions counted 30 trials. So, every participant performed 76 saccades in total. In the strict instructions condition, participants were told to hit the zero key of the numeric keyboard at a rate of one hit per second. In the lenient instructions condition, participants were told to hit the zero key of the numeric keyboard at a rate of about 1 hit per second, but it was stressed that small deviations were not that important. 3.2. Results Fig. 2 shows the mean latencies as a function of load and instructions for both prosaccades and antisaccades. The data-analysis revealed that for the prosaccades, the eect of load was not signi®cant (F(1, 14) 1.8, P > 0.05) nor was the eect of instruction (F(1, 14) 2.2, P > 0.05). The interaction was not signi®cant either F < 1. For the antisaccades, the eects of load and instruction were signi®cant (F(1, 14) 6.3 P < 0.05 and F(1, 14) 4.6, P < 0.05, respectively). Their interaction fell short from signi®cance (F(1, 14) 3.7, P 0.076). However, analysis of the simple main eects revealed clearly that the eect of ®xed tapping under lenient instructions was not signi®cant F < 1. Under strict instructions the eect was signi®cant (F(1, 7) 6.1, P < 0.05). The interaction of saccade task, instruction and load did not reach signi®cance (F(1, 28) 1.2, P > 0.05). Table 2 presents the mean number of errors for the prosaccade and the antisaccade task under dierent instructions and under load and noload conditions. None of the eects were signi®cant here (all Fs < 1).
78
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
Fig. 2. Latencies in ms for prosaccades (PS) and antisaccades (AS) as a function of load and task instructions in Experiment 2. Table 2
Mean number of errors per conditiona in Experiment 2 Lenient
Prosaccades Antisaccades a
Strict
Control
Fixed tapping
Control
Fixed tapping
0.13 (0.33) 10.00 (7.11)
0.13 (0.33) 11.75 (5.29)
0.38 (0.70) 9.50 (6.38)
0.25 (0.66) 9.63 (5.91)
Standard deviations are given in parentheses.
3.3. Discussion The results of this experiment suggest that the ®xed tapping eect is not motor based, because under lenient instructions the eect disappears both in prosaccades and antisaccades. Under strict instructions the eect of tapping is clearly present in the antisaccades, but absent in the prosaccades. This result advocates an explanation based on cognitive control of ®xed tapping when participants are given strict instructions. Although this experiment partially clari®es the results of Experiment 1, we cannot overlook a new problematic issue that arose. When we compare the ®rst two experiments, prosaccade latencies under strict ®xed tapping instructions were signi®cantly in¯uenced by the dual-task load in Experiment 1, but not in Experiment 2. This problem is addressed in the next section. 4. Additional analysis A closer look at the data reveals that the mean prosaccade latencies in the strict ®xed tapping condition are higher in Experiment 1 than in Experiment 2. An
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
79
additional analysis on the mean latencies of the control condition per participant revealed that some participants seem to be slower (mean control condition latencies longer than 200 ms). In Experiment 1 the number of slow participants was higher (5 out of 8) than in Experiment 2 (1 out of 8). Therefore a combined analysis on the latencies of the participants of both experiments was performed. Multivariate analyses showed that the dierence between the control and the strict ®xed tapping condition was signi®cant for the slow group (Mcontrol 229 ms, Mfixed 245 ms) (F(1, 5) 8.9, P < 0.05). For the fast group (Mcontrol 182 ms, Mfixed 190 ms), however, this eect was not signi®cant (F(1, 9) 1.8, P 0.21). This suggests that participants who are generally slower are more vulnerable to dual-task interference. The dierence in the proportion of slow participants for both experiments could explain why ®ndings look inconsistent at ®rst sight. Why are the slower participants more aected? A possible hypothesis is that they control the prosaccade execution rather than perform them automatically (which is in fact faster, i.e. a re¯exive reaction). The data of course cannot prove this hypothesis, but here is a link to the question which was already pointed out in the discussion of Experiment 1. The original working hypothesis was that interference would occur with saccades that require the inhibition of a prepotent response towards the stimulus. In order to make our hypothesis acceptable, Experiment 3 was designed to show that mere controlled execution, without inhibition, is enough to obtain interference eects.
5. Experiment 3 The above-mentioned dierence between automatically and controlled executed prosaccades is related to the dierence between exogenous and endogenous saccades, where the exogenous ones are triggered by a stimulus, as was the case in Experiment 1 and 2, and the endogenous saccades are performed without using a peripheral stimulus. In this experiment, two tasks were used: an exogenous prosaccade task, and a saccade task in which participants can only perform the saccades in a controlled manner (no peripheral stimulus) but do not need to inhibit prepotent saccades (endogenous saccade task). Both tasks were combined with the RIG task. We expected that a load on the executive functions would have a pronounced eect on the endogenous saccades, and would not, or at least only slightly, disrupt the prosaccades, due to trials in which the execution was controlled. 5.1. Method 5.1.1. Participants and design Sixteen psychology students (11 female, 5 male) of the University of Gent participated for course requirements and credit. They volunteered for this particular experiment. All participants had normal vision, and none of them had participated in Experiment 1 or 2.
80
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
A 2 (Saccade task: exogenous or endogenous prossaccade task) ´ 2 (Load: control, random time interval generation) factorial design was used with all variables used as within-subjects variables. As Experiment 2 has shown that ®xed tapping task varies with task instructions, we preferred to use the RIG task in this experiment. 5.1.2. Materials and procedure Method of measurement and materials were identical to those used in Experiments 1 and 2, except for the task stimuli. For the exogenous prosaccade task, the size of the peripheral stimulus was enlarged (side of the square 1 cm or 0.95°). The ®xation point was now a small cross (2 2 mm or 0.18°). The larger peripheral stimuli and the smaller ®xation point were used to increase the exogenous triggering of the prosaccades. For the endogenous saccade task, this cross was also used as ®xation point, which was replaced by an arrow pointing to the left or right, indicating the direction in which the saccade had to be made. In order to get saccades of a comparable amplitude, two very small squares (side 1 mm or 0.09°) remained visible throughout the endogenous saccade task conditions, both to the left and the right of the ®xation cross. The same distance between these squares and the ®xation point was used as in the previous experiments. Participants were instructed to make a saccade to the left or right square, depending on the direction of the arrow. All participants received the conditions in the same order: prosaccades, RIG single task, prosaccades and RIG, endogenous saccades and endogenous saccades combined with RIG. Every saccade condition consisted of 30 saccade stimuli. Both saccade tasks were practised during 8 trials. So each participant executed 136 saccades. In the RIG single task condition, participants performed the RIG task alone during a period equal to the length of one saccade condition (about 3 min). 5.2. Results After analysing the tapping data, one participant apparently did not comply with the secondary task instructions and his data were excluded from all further analyses. The error rate in all conditions and for all participants was less than 0.5% and appeared to be an insensitive measurement in this experiment. Therefore only the analysis on the latencies is given. Fig. 3 summarises the main results. Multivariate analyses on the data revealed that the endogenous saccades (M 352 ms) were performed signi®cantly slower than the prosaccades (M 199 ms) (F(1, 14) 83.6, P < 0.001). The main eect of RIG load was signi®cant (F(1, 14) 37.4, P < 0.001). However, the RIG eect was much larger for the endogenous saccades (F(1, 14) 14.2, P < 0.01). Analysing both the saccade groups separately though, revealed that the eect of load was signi®cant for both saccade tasks, F(1, 14) 7.9, P < 0.05 for the prosaccades and F(1, 14) 25.8 with P < 0.001 for the endogenous saccades. 5.3. Discussion This experiment shows that interference is not restricted to saccade tasks in which a prepotent response must be inhibited. In the endogenous saccade task participants
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
81
Fig. 3. Latencies in ms for prosaccades (PS) and endogenous saccades (ES) as a function of load in Experiment 3.
could only generate saccades in a controlled way, due to the lack of peripheral stimuli. In the exogenous saccade task correct saccades can be generated automatically, but some participants, or perhaps even all participants on some trials, could have controlled their eye movements. Although a slight interference eect was found on the exogenous prosaccades, the eect of the RIG task was much larger on the endogenously generated saccades. This eect supports the hypothesis that the performing of a prosaccade in a controlled fashion increases the vulnerability of dualtask interference. The manipulation of stimulus presentation (i.e. smaller ®xation point, larger peripheral stimulus) seems to decrease the prosaccade latencies, but apparently cannot prevent that at least some trials (the controlled ones?) were in¯uenced by the RIG load. This seems to suggest that one cannot assume prosaccades to be of a completely automatic nature when studying them in an arti®cial setting like an experiment.
6. General discussion In line with our hypothesis, the ®rst experiment revealed that an executive interference task increased the latencies and the number of errors in the antisaccade task. But, although the antisaccades are more prone to disruption, an eect on the prosaccade latencies was also found. Moreover, a closer inspection of the interaction of saccade task and load showed that the interference eect was equally strong for both the RIG task and the ®xed tapping task in the two saccade tasks. These unexpected ®ndings were addressed in Experiments 2 and 3. Experiment 2 compared the eects of strict and lenient ®xed tapping. It appeared that under more lenient instructions for the ®xed tapping task, there was no eect on either prosaccades or antisaccades. Under more strict instructions, however, there was an eect, but on antisaccades only. As both ®xed tapping tasks require the same
82
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
motor actions, while antisaccades require controlled processing, it seems that the observed interference eect is of a cognitive nature. Interestingly, Experiment 1 has shown that the load of the secondary tasks had an eect on both saccade tasks. The working hypothesis was that prosaccades, in view of their supposed automatic nature, would be insensitive to executive interference. The interference found on the prosaccades therefore is unexpected, if we assume that interference occurs with saccades that need the inhibition of a prepotent response. However, it is possible that the component ``controlled'' is equally or more important in the execution of saccades. Considering the rather high latencies of some of the participants on the prosaccade task in the control condition, it could be that some participants controlled the execution of some of the prosaccades, generating these saccades through the FEF rather than through the CS. Convinced that a planned execution of the prosaccades will increase their performance, they not only increase the latencies on the prosaccade task, but they may also have made this task vulnerable to dual-task interference. The additional analysis revealed that only the latencies of the slower group of participants were in¯uenced by the strict ®xed tapping load. These data only suggest and do not really prove the idea of controlled execution. Presuming this idea is correct, further support for this hypothesis was collected by showing that mere planned execution of a saccade can cause cognitive interference eects. Using an endogenous prosaccade task, Experiment 3 showed that the fact that a saccade is performed in a controlled fashion is indeed a source of interference. Moreover, the amount of interference is comparable to the interference found on the antisaccade task. So that at least in these experiments, no evidence was found for interference due to inhibition of a prepotent saccade over and above the interference eect found on planned saccades. In the experiment of Roberts et al. (1994) with mental arithmetic as secondary task, no interference on the prosaccades was found. A possible explanation is based on the fact that mental arithmetic loads more heavily on the executive functions than our tapping tasks. As a result more executive control is needed for this secondary task and less executive capacity is left for the prosaccade task which is then performed with less control, resulting in more re¯exive prosaccades. Support for this idea can be found in an experiment of Crevits et al. (1998). Here, prefrontal patients (excluding FEF lesions) executed the prosaccade and the antisaccade task both as a single task and in combination with the RIG task. Compared to a control group of healthy participants, only the antisaccades were impaired, but not the prosaccades. The latter were not in¯uenced by the RIG task. Here appears a similarity to the heavy load of mental arithmetic, the frontal lesion leaves the participants little capacity for controlled execution of the prosaccades, and consequently these saccades are more likely to be performed in the most natural way, i.e. automatically. A study by Henik, Rafal and Rhodes (1994) reports converging evidence for the importance of the controlled generation of saccades and executive interference: patients with frontal lesions performed worse on an endogenous saccade task, similar to the one we used in Experiment 3. Pashler, Carrier and Homan (1993) also found dual-task interference on an endogenous prosaccade task using a
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
83
psychological refractory period (PRP) paradigm. However, only the latencies of the saccade tasks that had to be executed in a controlled manner, showed a PRP eect, suggesting that in these experiments, the load of the PRP trials was so high that participants generated the prosaccades without control whenever possible. For both the saccade tasks and the secondary tapping tasks, the original focus was on inhibition (of a prepotent saccade, of regular time interval sequences, i.e. the random instruction) as a source of interference. In sum, our results suggest that inhibition here had no additional importance, and the controlled processing of both saccades and tapping behaviour could be the origin of the observed cognitive interference. This can explain the equal eects of RIG and strict ®xed tapping and the eect on the exogenous prosaccades. This eect is of course smaller than the one observed with antisaccades and endogenous saccades, because in the latter tasks all saccades have to be planned. In the exogenous prosaccade task on the contrary, only a proportion of the trials could have been planned, and this proportion varied across subjects; in the remaining trials the saccades could be executed automatically. By using secondary tasks that do not rely so heavily on executive control as the mental arithmetic task used by Roberts et al. (1994), it appears that healthy participants under secondary task load are not completely comparable with, for example, prefrontal patients. One methodological note, and a suggestion for further research, is in place here. We believe that the dual-task paradigm is perhaps a research technique at too macroscopic a level. Saccades are known to be carried out rather quickly: depending on circumstances and task requirements, saccades vary between 80 and 450 ms. Tapping behaviour was instructed to be at a pace of about one tap per second. On some trials, the decision to tap could have occurred 800 ms before the programming of the eye movement, on other trials it could almost have coincided with it. For this reason, we believe that designs which allow a trial per trial analysis of the eect of temporal overlap could oer a more microscopic and ecient methodology to investigate the interaction of eye movement control and other executive functions. One possibility of this trial by trial method is to use response-response compatibility as a factor aecting saccade performance in dual-task conditions. Currently, we use this approach to investigate dierences in action-planning for automatic and controlled saccades.
Acknowledgements The authors are indebted to Marc Brysbaert, Gino De Vooght, Gezinus Wolters and two anonymous reviewers for their valuable comments on an earlier version of this paper and to Eva Kemps for checking and improving the English. Part of the research reported in this paper was supported by grant G0012.97 from the Fund for Scienti®c Research-Flanders to Andre Vandierendonck and a doctoral scholarship from the Research Fund of the University of Gent to Els Stuyven (BOF no. 011D0896).
84
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
References Baddeley, A. D. (1966). The capacity for generating information by randomisation. The Quarterly Journal of Experimental Psychology, 18, 119±129. Baddeley, A. (1986). Working memory. Oxford University Press: Oxford. Baddeley, A. (1990). Human memory: theory and practice. London: Erlbaum. Baddeley, A. D., & Hitch, J. G. (1974). Working memory. In G. H. Bower, The psychology of learning and motivation, vol. 8. (pp. 47±89). New York: Academic Press. Baddeley, A. D., Lewis, V., Eldridge, M., & Thomson, N. (1984). Attention and retrieval from long-term memory. Journal of Experimental Psychology (General), 113, 518±540. Claeys, K., Crevits, L., Stuyven, E., Van der Goten, K., Depuydt, C., & Vandierendonck, A. (1999). Parallel visual and memory processes. Documenta Opthalmologica, 95, 349±358. Crevits, L., & De Ridder, K. (1997). Disturbed striato-prefrontal mediated visual behaviour in moderate to severe parkinsonian patients. Journal of Neurology Neurosurgery Psychiatry, 63, 296±299. Crevits, L., Claeys, K., Stuyven, E., Depuydt, C., Van der Goten, K., & Vandierendonck, A. (submitted). Saccadic behaviour and working memory in prefrontal patients, submitted. Dalrymple-Alford, J. C., Kalders, A. S., Jones, R. D., & Watson, R. W. (1994). A central executive de®cit in patients with Parkinson's disease. Journal of Neurology Neurosurgery and Psychiatry, 57, 360± 367. De Rammelaere, S., Stuyven, E., & Vandierendonck, A. (1999). The contribution of working memory resources in the veri®cation of simple mental arithmetic sums. Psychological Research, 62, 72±77. Fukushima, J., Fukushima, K., Chiba, T., Tanaka, S., Yamashita, I., & Kato, M. (1988). Disturbances of voluntary control of saccadic eye movements in schizophrenic patients. Biological Psychiatry, 23, 670± 677. Fukushima, J., Fukushima, K., Miysaka, K., & Yamashita, I. (1994). Voluntary control of saccadic eye movements in patients with frontal cortical lesions and parkinsonian patients in comparison with that in schizophrenics. Biological Psychiatry, 36, 21±30. Fukushima, J., Fukushima, K., Morita, N., & Yamashita, I. (1990). Further analysis of the control of voluntary saccadic eye movements in schizophrenic patients. Biological Psychiatry, 28, 943±958. Gathercole, S. E. (1994). Neuropsychology and working memory: a review. Neuropsychology, 8, 494±505. Gilhooly, K. J., Logie, R. H., Wetherick, N. E., & Wynn, V. (1993). Working memory and strategies in syllogistic-reasoning tasks. Memory and Cognition, 21, 115±124. Glanzer, M., Dorfman, D., & Kaplan, B. (1981). Short-term storage in the processing of text. Journal of Verbal Learning and Verbal Behavior, 20, 656±670. Greenwald, A. G., & Shulman, H. G. (1973). On doing two things at once: II Elimination of the psychological refractory period eect. Journal of Experimental Psychology, 101, 70±76. Guitton, D., Buchtel, H. A., & Douglas, R. M. (1985). Frontal lobe lesions in man cause diculties in suppressing re¯exive glances and in generating goal-directed saccades. Experimental Brain Research, 58, 455±472. Hallet, P. E., & Adams, B. D. (1980). The predictability of saccadic latency in a novel voluntary oculomotor task. Vision Research, 20, 329±339. Hallet, P. E. (1978). Primary and secondary saccades to goals de®ned by instructions. Vision Research, 18, 1279±1296. Henik, A., Rafal, R., & Rhodes, D. (1994). Endogenously generated and visually guided saccades after lesions of the human frontal eye ®elds. Journal of Cognitive Neuroscience, 6, 400±411. Kitagawa, M., Fukushima, J., & Tashiro, K. (1994). Relationship between antisaccades and the clinical symptoms in Parkinson's disease. Neurology, 44, 2285±2289. Logie, R. H., Gilhooly, K. J., & Wynn, V. (1994). Counting on working memory in arithmetic problem solving. Memory and Cognition, 22, 395±410. Martein, R., Kemps, E., & Vandierendonck, A. (1999). The role of working memory in a double span task. Psychologica Belgica, 39, 15±27. McCall, R. B., & Appelbaum, M. I. (1973). Bias in the analysis of repeated measure designs: some alternative approaches. Child Development, 44, 401±415.
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
85
Pashler, H., Carrier, M., & Homan, J. (1993). Saccadic eye movements and dual-task interference. The Quarterly Journal of Experimental Psychology, 46A, 51±82. Rapee, R. M. (1993). The utilisation of working memory by worry. Behaviour Research Therapy, 31, 617±620. Roberts, R. J., Hager, L. D., & Heron, C. (1994). Prefrontal cognitive processes: working memory and inhibition in the antisaccade task. Journal of Experimental Psychology: General, 123, 374±393. Rosse, R. B., Schwartz, B. L., Kim, S. Y., & Deutsch, J. I. (1993). Correlation between antisaccade and wisconsin card sorting test performance in schizofrenia. American Journal of Psychiatry, 150, 333±335. Rusted, J. M. (1988). Dissociative eect scopolamine on working memory in healthy young volunteers. Psychopharmacology, 96, 487±492. Stuyven, E., & Van der Goten, K. (1995). Stimulus independent thoughts and working memory: the role of the central executive. Psychologica Belgica, 35, 241±251. Vandierendonck, A., De Vooght, G., & Van der Goten, K. (1998). Does random time interval generation interfere with working memory executive functions? European Journal of Cognitive Psychology, 10, 413±442.
www.elsevier.com/locate/actpsy
The eect of cognitive load on saccadic eye movements Els Stuyven a
a,*
, Koen Van der Goten a, Andre Vandierendonck a, Kristl Claeys b, Luc Crevits b
Department of Experimental Psychology, University of Gent, Henri Dunantlaan 2, B-9000 Gent, Belgium b Department of Neurology, University of Gent, Belgium Received 22 February 1999; received in revised form 20 September 1999; accepted 20 September 1999
Abstract The present study tested the hypothesis that, unlike prosaccades, antisaccades require controlled processing, due to the prepotent response that needs to be inhibited. The eect of the Random time Interval Generation (RIG) task (Vandierendonck, A., De Vooght, G., & Van der Goten, K. (1998). European Journal of Cognitive Psychology, 10, 413±444) on these saccade latencies and errors was studied. This task has the advantage that it loads executive processes, with only minimal interference with verbal or visuo-spatial components. A ®rst experiment compared saccade performance within the prosaccade and the antisaccade task, executed alone and in combination with the RIG task and ®xed tapping (added to exclude possible motor component interference explanations). A second experiment investigated the in¯uence of task characteristics on the eects found. Although it was shown that antisaccades are more prone to interference of an executive interference task, it seems that prosaccades are also vulnerable. Interference on prosaccades could originate from a controlled execution of these saccades. A third experiment con®rmed that endogenously generated prosaccades are susceptible to dual-task interference and showed that controlled saccade execution, without the need to inhibitit a prepotent response, is sucient to produce interference. Ó 2000 Elsevier Science B.V. All rights reserved. PsycINFO classi®cation: 2323; 2340; 2346
*
Corresponding author. Tel.: +32-(0)9-2646406; fax: +32-(0)9-2646496. E-mail address: [email protected] (E. Stuyven).
0001-6918/00/$ - see front matter Ó 2000 Elsevier Science B.V. All rights reserved. PII: S 0 0 0 1 - 6 9 1 8 ( 9 9 ) 0 0 0 5 4 - 2
70
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
1. Introduction Recent evidence suggests that eye movements towards a laterally presented stimulus (prosaccades) are performed automatically, whereas eye movements in a direction opposed to the normal saccadic direction (antisaccades) are performed under attentional control (Roberts, Hager & Heron, 1994). The antisaccade task was developed by Hallet (1978; Hallet & Adams, 1980) to investigate the mechanisms that underly the generation of automatic and voluntary eye movements. He found that antisaccadic eye movements typically have longer latencies (time between stimulus onset and the start of the eye movement), that inaccurate re¯exive saccades towards the cue sometimes occur, and that practice does not result in an improvement of antisaccade performance. 1.1. Antisaccades and frontal lobes Guitton, Buchtel and Douglas (1985) used the antisaccade task to investigate de®cient inhibitory control in patients with prefrontal lesions. The Frontal Eye Field (FEF, area 8) and the Colliculus Superior (CS) are two systems involved in the generation of saccadic eye movements. The CS is supposed to mediate automatic, re¯exive movements, whereas the FEF is assumed to be involved in higher level control of saccadic eye movements. The results of the frontal patients were compared with those of a control group and a group with lesions in the temporal lobe. No dierences between the groups were found in latencies or errors of prosaccades. In the antisaccade task, however, frontal patients made more re¯exive saccades (i.e., incorrect saccades towards the stimulus). Moreover, some frontal patients only performed a re¯exive saccade with stimulus presentation contralateral to the damaged hemisphere. The results suggest that the damaged structures are involved in the generation of the antisaccades. This strengthens the idea that the prefrontal cortex contributes to the generation of complex saccadic eye movement behaviour. The function of these lobes would then be the inhibition of re¯exive oculomotor activity and the triggering of voluntary eye movements. Other authors used the antisaccade task in populations where prefrontal dysfunctioning is hypothesised (e.g., schizophrenia, ADHD). Fukushima et al. (1988) studied prosaccade and antisaccade performance in schizophrenic patients. Compared to healthy control participants, no dierences were observed in the prosaccade condition. However, half of the schizophrenic patients made more incorrect antisaccades, and the correct antisaccades had longer latencies. Fukushima, Fukushima, Morita and Yamashita (1990) presented schizophrenic patients with a nosaccade task and a memory saccade task. In the nosaccade task, participants ®xate a central point in spite of the presentation of a cue. In the memory saccade task, a cue is presented while the ®xation point is still present. Only after the removal of the ®xation point a saccade must be made to the point where the cue was previously presented. Schizophrenic patients made a high number of saccades in the nosaccade task, whereas the healthy controls made no mistakes. Moreover, the memory saccade latencies were longer in schizophrenic patients than in the controls.
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
71
In view of this evidence it seems that schizophrenic patients, much like prefrontal patients, have diculty inhibiting re¯exive saccades and executing voluntary saccades when the target position is known but not visible. The connection with the frontal lobes is also supported by the work of Rosse, Schwartz, Kim and Deutsch (1993) who found a signi®cant correlation between performance on the antisaccade task and the Wisconsin Card Sorting Task in schizophrenic patients. Crevits and De Ridder (1997) studied antisaccades and memory saccades in a large group of Parkinson's patients. They demonstrated severe abnormalities suggesting a dysfunction in the striato-prefrontal lobe. However, previous reports in smaller groups of less severely aected patients could not show consistent abnormalities (Kitagawa, Fukushima & Tashiro, 1994; Fukushima, Fukushima, Miysaka & Yamashita, 1994). 1.2. Antisaccades and working memory The multi-component working memory model proposed by Baddeley and Hitch (1974) and Baddeley (1986, 1990), has proven to be a very fruitful conceptualisation of short-term memory. It has earned its merits in various areas of cognitive psychology such as reading comprehension, mental arithmetic, imagery, reading development, problem solving and others. Moreover, the model has demonstrated its usefulness in the study of patients with neurological de®cits. It has, for instance, been found that interference with attentional control in dual-tasks, results in patterns that are similar to those found in patients with prefrontal lesions (Baddeley, 1990). Roberts et al. (1994) investigated the relationship between antisaccades and working memory. Like Gathercole (1994), these researchers assume two important functions of the prefrontal cortex: (1) the short-term retention of information in order to organise future actions, and (2) the inhibition of prepotent responses. In addition, the authors assume that a temporary increase in working memory load can result in patterns of errors similar to patterns of permanent dysfunction of working memory, like those observed in prefrontal patients. Roberts et al. (1994) compared the single-task and dual-task performance in a prosaccade and an antisaccade task. The secondary tasks used were mental arithmetic, a shadowing task (i.e. repeat a given number), or a repeat task (i.e. always repeat the same number). Logie, Gilhooly and Wynn (1994) have already shown that mental arithmetic loads the central executive and the phonological loop. In this experiment Roberts et al. (1994), used only healthy participants. In the prosaccade task, almost no errors were made and mental arithmetic did not aect saccade latency. In the antisaccade task, however, more directional errors (re¯exive saccades towards the stimulus) occurred and saccade latencies increased, much like in the prefrontal patients of Guitton et al. (1985). These results suggest at least a behavioural similarity between frontal patients and non-patients under heavy working memory load. Roberts et al. (1994) assumed that as long as working memory is able to intervene in response preparation, alternative responses can be inhibited. The CS, which takes part in re¯exive saccades receives input directly (FEF±CS) and indirectly (FEFcaudate nucleus-substantia nigra-CS) from the dorsolateral prefrontal cortex. The working memory system can inhibit the CS, for instance to prevent re¯exive saccades
72
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
in the antisaccade task and it may also be involved in initiating a planned saccade. However, the more working memory resources are involved in other activities, the less resources are available for inhibitory control and planning, and the more probable it will be for directional errors to occur in the antisaccade task. In order to show unequivocally that the central executive, rather than the entire working memory system, takes part in antisaccade preparation, the concurrent task should be speci®c: it should interfere with the central executive without loading the slave systems. The literature contains a large number of tasks that interfere with the central executive: e.g., random number generation (Baddeley, 1966), random digit generation (e.g., Gilhooly, Logie, Wetherick & Wynn, 1993), tapping a random key pattern (Rapee, 1993), avoidance of automaticity in a verbal reaction task (Greenwald & Shulman, 1973), random pursuit tracking (Dalrymple-Alford, Kalders, Jones & Watson, 1994), free recall (Rusted, 1988), counting backwards (e.g., Glanzer, Dorfman & Kaplan, 1981), and generation of category members (e.g., Baddeley, Lewis, Eldridge & Thomson, 1984). All these tasks have the drawback of also interfering with either the visuo-spatial or the phonological slave subsystem, so that clear-cut conclusions are not easy to obtain. Vandierendonck, De Vooght and Van der Goten (1998) have developed and tested a central executive task that is allegedly pure in the sense that it does not load signi®cantly either of the known slave systems. In this Random time Interval Generation (RIG) task, participants are to tap a randomly spaced sequence of time intervals on a key so as to produce a completely unpredictable ``rhythm''. In order to tap randomly, automaticity has to be prevented, and so this task requires cognitive control, while there are no logical or empirical reasons to assume that it interferes with the slave systems. This task has already been shown to aect span, supraspan (Vandierendonck et al., 1998), stimulus independent thoughts (Stuyven & Van der Goten, 1995), a dual content location span task (Martein, Kemps & Vandierendonck, 1999) and simple mental arithmetic (De Rammelaere, Stuyven & Vandierendonck, 1999). The reported experiments were inspired by the working memory model. However, because the concept of the ``central executive'' and its until recently presumed unitary nature stands to discussion, we prefer to use the more neutral term ``executive functions''. 2. Experiment 1 The eect of the RIG task on saccades was ®rst studied in Claeys et al. (1999). Prosaccades and antisaccades were executed in a single- and a dual-task (i.e. in combination with RIG) situation. On the basis of the hypothesis that prosaccades are automatic, the authors expected no interference of the secondary task with prosaccade performance. However, it turned out that both prosaccade and antisaccade latencies are in¯uenced by the RIG load and that there is no kind of interaction of saccade task and load. The question can be raised whether a possible eect of RIG on the saccades could be due to a central cognitive component, a motor component, or simply to the fact that two tasks have to be performed at the same time. A central component would involve the cognitive control over the actions to be
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
73
performed (i.e., the inhibition of a re¯exive saccade, the start of a saccade in the opposite direction of the stimulus, the creation of a random series of time intervals). A possible motor interference would re¯ect the diculty of performing two motorbased actions at the same time (i.e., eye movement and ®nger movement). In order to address these possibilities, in Experiment 1 a ®xed tapping task was included as an additional control condition for RIG as it requires the same motor actions, but assumably does not call on executive functions. In this case the output does not have to be monitored for randomness, and inhibition of familiar patterns is not needed. 2.1. Method 2.1.1. Participants and design Sixteen ®rst-year psychology students (seven female, nine male) of the University of Gent participated for course requirements and credit. They volunteered for this particular experiment. All participants had normal vision. A 2 (Saccade task: prosaccade or antisaccade) ´ 2 (2 Order of presentation) ´ 3 (Load: control, ®xed tapping, RIG) mixed factorial design was used in which the saccade task was a between-subjects and load a within-subjects variable. Participants were randomly assigned to the saccade groups and within this group to either of two randomly chosen order conditions (Fixed-Control-RIG and Control-RIG-Fixed). 2.1.2. Materials and procedure Eye movements were registered by means of an infra-red based eye tracking system, the Ober2 system. Two IBM-compatible PC's connected by means of a parallel cable were used in this experiment in order to synchronise eye movement registration and the secondary task. The program to present the visual stimuli and register the eye movements was written in the Nystr om language, which is part of the eye tracking software, and was used in every condition. Participants were seated at a distance of 600 mm from a 14 inch. screen. The sampling frequency was 300 Hz, so that a measurement of eye position was taken every 3.33 ms. Only the movements of the dominant eye were registered. In the centre of the screen a white square (4 4 mm or visual angle 0.38°) was used as the ®xation point. As in the experiments of Guitton et al. (1985), after 1.5±3 s, the square disappeared. At the same time another square of the same size appeared 70 mm to the left or to the right (visual angle 6.7°). After 3±5 s, the screen was cleared and the central ®xation square reappeared. The secondary RIG task was monitored by the second PC. Participants were instructed to tap an unpredictable sequence on the zero key of the numeric keypad of the computer at an average rate of one keystroke per second. It was emphasised that repetition of a pattern was not allowed. These tap sequences were registered from the moment eye movement registration was started until the program for eye movement registration sent a stop signal (one condition lasted about 3 min). For the ®xed tapping task, participants were instructed to hit the zero key of the numeric keyboard at a ®xed rate of one per second. As in the RIG task, tapping behaviour was registered.
74
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
Performance was measured in terms of latencies and errors of saccades. Reaction time is de®ned as the time interval between the appearance of the stimulus and the onset of the eye movement. Possible errors are: (1) saccades in the wrong direction; (2) unperformed saccades; (3) re¯exive saccades in the antisaccade task. Anticipatory saccades (latencies less than 100 ms or started before stimulus onset) were not counted as errors but were removed from the analysis (about 1% of the trials). After the instructions were given, the participant put on the infrared goggles of the Ober2 system and the room was darkened. At this moment, a calibration phase was started in order to adjust a number of parameters of the eye tracking apparatus (e.g., illuminance). Participants passed two calibration sessions of eight trials in which the prosaccade or antisaccade task was practised. There was also a practice session for both secondary tasks. After calibration, the three conditions were run. Every condition consisted of 30 trials, so that including calibration, every participant performed 106 saccades. 2.2. Results The data were analysed by means of a multivariate analysis. The eects of the between-subjects and the within-subjects variables and their interactions were estimated by means of contrasts between the dependent variables: the mean latencies per participant under each of the three load conditions. For the accuracy analysis the average number of errors per condition was the dependent variable. This procedure conforms to the suggestions of McCall and Appelbaum (1973) for a correct dataanalysis of repeated measures designs. 1 2.2.1. Eect of order The order of the conditions had no eect on the latencies or on the number of errors. The interactions with saccade task and load were not signi®cant either (all Fs < 1). 2.2.2. Latencies Fig. 1 displays the mean latencies for the prosaccades and the antisaccades under control, ®xed tapping, and RIG conditions. The main eect of saccade task was signi®cant with prosaccades performed faster (M 228 ms) than antisaccades (M 325 ms) (F(1, 10) 9.5, P < 0.05). The main eect of load was also signi®cant (Mcontrol 246, Mfixed 288, MRIG 295) (F(2, 9) 61.2, P < 0.001). The control condition diered from both the RIG and the ®xed tapping condition (F(1, 10) 41.8, P < 0.01 and F(1, 10) 14.9, P < 0.01, respectively). The dierence between the RIG and the ®xed condition, however, was not signi®cant F < 1. Furthermore, as predicted, the saccade task ´ load interaction was signi®cant 1
The data of the random and ®xed tapping tasks were analysed according to the method described by Vandierendonck et al. (1998). None of the experiments reported trade-os with the performance on the saccade tasks, hence, these analyses will not be discussed further.
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
75
Fig. 1. Latencies in ms for prosaccades (PS) and antisaccades (AS) as a function of secondary task conditions in Experiment 1.
(F(2, 9) 27.4, P < 0.001). Decomposition of this interaction revealed that the RIG versus control contrast interacted with the saccade tasks (F(1, 10) 6.2, P < 0.05). Fig. 1 clearly shows that the dierence between the control and the RIG condition was smaller (but signi®cant) in the prosaccade than in the antisaccade task. It is clear from this ®gure that the interaction of saccade task and the ®xed tapping versus control contrast was also signi®cant (F(1, 10) 19.5, P < 0.01). But contrary to our expectation, the interaction of saccade task with the random versus ®xed tapping contrast was not found F < 1. This pattern of results indicates that the ®xed tapping and the RIG task had comparable eects on the eye movement latencies. We will return to this in the discussion. Despite the interaction of saccade task and load, showing that antisaccades were more profoundly aected by the interference than prosaccades, a closer inspection revealed that in both the prosaccade and the antisaccade task, the eect of load was signi®cant (F(1, 5) 7.7, P < 0.05 and F(1, 5) 64.9, P < 0.001, respectively). 2.2.3. Errors Table 1 shows the average number of errors in the prosaccade and the antisaccade task under the dierent secondary task conditions. More errors occurred in antisaccades (M 5.29) than in prosaccades (M 0.08). F(1, 10) 15.9, P < 0.01. The Table 1
Mean number of errors per conditiona in Experiment 1
Prosaccade Antisaccade a
Control
Fixed tapping
RIG
0.00 (0.00) 3.88 (3.95)
0.25 (0.43) 4.50 (4.39)
0.00 (0.00) 7.50 (6.69)
Standard deviations are given in parentheses (RIG: random time interval generation task).
76
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
eect of load was also signi®cant (Mcontrol 1.31, Mfixed 2.38, MRIG 3.75) (F(1, 10) 7.4, P < 0.05), as was the task ´ load interaction (F(2, 9) 4.4, P < 0.05). This interaction was in line with the hypothesis: the dierence between RIG and control was larger in the antisaccade than in the prosaccade task (F(1, 10) 9.1, P < 0.05), as was the dierence between RIG and ®xed tapping (F(1, 10) 7.6, P < 0.05). Moreover, the interaction of saccade task with the contrast between ®xed tapping and control was not reliable (F(1, 10) 1.7, P > 0.20). 2.3. Discussion The results partially con®rm our hypothesis; antisaccade performance was aected more by the secondary task load than latencies and errors in the prosaccade task. However, two unexpected ®ndings arose from this experiment. First, although the interaction of saccade task and load was signi®cant, prosaccade latencies increased signi®cantly in the dual-task conditions. This is contrary to the hypothesis that prosaccades would be invulnerable, due to their automatic nature. One possible explanation is that participants actually executed their prosaccades in a controlled way in order to improve their performance. It could be that a controlled saccade execution makes these saccades vulnerable to interference. To support this idea, it is necessary to collect empirical evidence that the controlled execution of a saccade suces to produce interference with an executive task. Roberts et al. (1994) hypothesised that executive functions play a role in the planning of a saccade. However, by using the antisaccade task, they could only con®rm the interaction of the inhibition of a prepotent saccade and executive functions. Experiment 3 was designed to clarify this point. A second remarkable result is that there was no signi®cant dierence in the saccade latencies of the RIG and the ®xed tapping condition. It is possible that we observed motor interference rather than cognitive interference. In fact, the ®xed tapping task was included to control for motor interference, as the same motor actions are required without taxing executive functions. A cognitive interpretation of the eect remains possible, however. First, why would this result in a larger impairment of antisaccade performance if the interference is motor based? Second, the instructions given in the ®xed tapping condition were rather strict. Participants were told to hit the zero-key of the numeric keyboard every second. If participants took this task very seriously and monitored the time elapsed since the last key press, ®xed tapping might not be considered an automatic task, but a task which requires executive control for its accurate execution. 2 To test this hypothesis, Experiment 2 compared performance under strict instructions and more lenient instructions for the ®xed tapping task on both prosaccades and antisaccades under single- and dual-task conditions.
2
We are indebted to Gino De Vooght for this suggestion.
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
77
3. Experiment 2 3.1. Method 3.1.1. Participants and design Thirty two ®rst-year psychology students (26 female, 6 male) of the University of Gent participated for course requirements and credit. They volunteered for this particular experiment. All participants had normal vision, none of them had participated in Experiment 1. A 2 (Saccade task: prosaccade or antisaccade) ´ 2 (Load: control, ®xed tapping) ´ 2 (Instruction: strict, lenient) mixed factorial design was used with saccade task and instructions as between-subjects variables and with load as a within-subjects variable. Participants were randomly assigned to one of the four between-subjects cells. As Experiment 1 has shown that the condition order had no eect, only one order was used here. 3.1.2. Materials and procedure The materials were identical to those of Experiment 1. Every participant ®rst performed the control condition of the experiment. Both the control and the ®xed tapping condition were preceded by a practice session of eight saccades. The experimental sessions counted 30 trials. So, every participant performed 76 saccades in total. In the strict instructions condition, participants were told to hit the zero key of the numeric keyboard at a rate of one hit per second. In the lenient instructions condition, participants were told to hit the zero key of the numeric keyboard at a rate of about 1 hit per second, but it was stressed that small deviations were not that important. 3.2. Results Fig. 2 shows the mean latencies as a function of load and instructions for both prosaccades and antisaccades. The data-analysis revealed that for the prosaccades, the eect of load was not signi®cant (F(1, 14) 1.8, P > 0.05) nor was the eect of instruction (F(1, 14) 2.2, P > 0.05). The interaction was not signi®cant either F < 1. For the antisaccades, the eects of load and instruction were signi®cant (F(1, 14) 6.3 P < 0.05 and F(1, 14) 4.6, P < 0.05, respectively). Their interaction fell short from signi®cance (F(1, 14) 3.7, P 0.076). However, analysis of the simple main eects revealed clearly that the eect of ®xed tapping under lenient instructions was not signi®cant F < 1. Under strict instructions the eect was signi®cant (F(1, 7) 6.1, P < 0.05). The interaction of saccade task, instruction and load did not reach signi®cance (F(1, 28) 1.2, P > 0.05). Table 2 presents the mean number of errors for the prosaccade and the antisaccade task under dierent instructions and under load and noload conditions. None of the eects were signi®cant here (all Fs < 1).
78
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
Fig. 2. Latencies in ms for prosaccades (PS) and antisaccades (AS) as a function of load and task instructions in Experiment 2. Table 2
Mean number of errors per conditiona in Experiment 2 Lenient
Prosaccades Antisaccades a
Strict
Control
Fixed tapping
Control
Fixed tapping
0.13 (0.33) 10.00 (7.11)
0.13 (0.33) 11.75 (5.29)
0.38 (0.70) 9.50 (6.38)
0.25 (0.66) 9.63 (5.91)
Standard deviations are given in parentheses.
3.3. Discussion The results of this experiment suggest that the ®xed tapping eect is not motor based, because under lenient instructions the eect disappears both in prosaccades and antisaccades. Under strict instructions the eect of tapping is clearly present in the antisaccades, but absent in the prosaccades. This result advocates an explanation based on cognitive control of ®xed tapping when participants are given strict instructions. Although this experiment partially clari®es the results of Experiment 1, we cannot overlook a new problematic issue that arose. When we compare the ®rst two experiments, prosaccade latencies under strict ®xed tapping instructions were signi®cantly in¯uenced by the dual-task load in Experiment 1, but not in Experiment 2. This problem is addressed in the next section. 4. Additional analysis A closer look at the data reveals that the mean prosaccade latencies in the strict ®xed tapping condition are higher in Experiment 1 than in Experiment 2. An
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
79
additional analysis on the mean latencies of the control condition per participant revealed that some participants seem to be slower (mean control condition latencies longer than 200 ms). In Experiment 1 the number of slow participants was higher (5 out of 8) than in Experiment 2 (1 out of 8). Therefore a combined analysis on the latencies of the participants of both experiments was performed. Multivariate analyses showed that the dierence between the control and the strict ®xed tapping condition was signi®cant for the slow group (Mcontrol 229 ms, Mfixed 245 ms) (F(1, 5) 8.9, P < 0.05). For the fast group (Mcontrol 182 ms, Mfixed 190 ms), however, this eect was not signi®cant (F(1, 9) 1.8, P 0.21). This suggests that participants who are generally slower are more vulnerable to dual-task interference. The dierence in the proportion of slow participants for both experiments could explain why ®ndings look inconsistent at ®rst sight. Why are the slower participants more aected? A possible hypothesis is that they control the prosaccade execution rather than perform them automatically (which is in fact faster, i.e. a re¯exive reaction). The data of course cannot prove this hypothesis, but here is a link to the question which was already pointed out in the discussion of Experiment 1. The original working hypothesis was that interference would occur with saccades that require the inhibition of a prepotent response towards the stimulus. In order to make our hypothesis acceptable, Experiment 3 was designed to show that mere controlled execution, without inhibition, is enough to obtain interference eects.
5. Experiment 3 The above-mentioned dierence between automatically and controlled executed prosaccades is related to the dierence between exogenous and endogenous saccades, where the exogenous ones are triggered by a stimulus, as was the case in Experiment 1 and 2, and the endogenous saccades are performed without using a peripheral stimulus. In this experiment, two tasks were used: an exogenous prosaccade task, and a saccade task in which participants can only perform the saccades in a controlled manner (no peripheral stimulus) but do not need to inhibit prepotent saccades (endogenous saccade task). Both tasks were combined with the RIG task. We expected that a load on the executive functions would have a pronounced eect on the endogenous saccades, and would not, or at least only slightly, disrupt the prosaccades, due to trials in which the execution was controlled. 5.1. Method 5.1.1. Participants and design Sixteen psychology students (11 female, 5 male) of the University of Gent participated for course requirements and credit. They volunteered for this particular experiment. All participants had normal vision, and none of them had participated in Experiment 1 or 2.
80
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
A 2 (Saccade task: exogenous or endogenous prossaccade task) ´ 2 (Load: control, random time interval generation) factorial design was used with all variables used as within-subjects variables. As Experiment 2 has shown that ®xed tapping task varies with task instructions, we preferred to use the RIG task in this experiment. 5.1.2. Materials and procedure Method of measurement and materials were identical to those used in Experiments 1 and 2, except for the task stimuli. For the exogenous prosaccade task, the size of the peripheral stimulus was enlarged (side of the square 1 cm or 0.95°). The ®xation point was now a small cross (2 2 mm or 0.18°). The larger peripheral stimuli and the smaller ®xation point were used to increase the exogenous triggering of the prosaccades. For the endogenous saccade task, this cross was also used as ®xation point, which was replaced by an arrow pointing to the left or right, indicating the direction in which the saccade had to be made. In order to get saccades of a comparable amplitude, two very small squares (side 1 mm or 0.09°) remained visible throughout the endogenous saccade task conditions, both to the left and the right of the ®xation cross. The same distance between these squares and the ®xation point was used as in the previous experiments. Participants were instructed to make a saccade to the left or right square, depending on the direction of the arrow. All participants received the conditions in the same order: prosaccades, RIG single task, prosaccades and RIG, endogenous saccades and endogenous saccades combined with RIG. Every saccade condition consisted of 30 saccade stimuli. Both saccade tasks were practised during 8 trials. So each participant executed 136 saccades. In the RIG single task condition, participants performed the RIG task alone during a period equal to the length of one saccade condition (about 3 min). 5.2. Results After analysing the tapping data, one participant apparently did not comply with the secondary task instructions and his data were excluded from all further analyses. The error rate in all conditions and for all participants was less than 0.5% and appeared to be an insensitive measurement in this experiment. Therefore only the analysis on the latencies is given. Fig. 3 summarises the main results. Multivariate analyses on the data revealed that the endogenous saccades (M 352 ms) were performed signi®cantly slower than the prosaccades (M 199 ms) (F(1, 14) 83.6, P < 0.001). The main eect of RIG load was signi®cant (F(1, 14) 37.4, P < 0.001). However, the RIG eect was much larger for the endogenous saccades (F(1, 14) 14.2, P < 0.01). Analysing both the saccade groups separately though, revealed that the eect of load was signi®cant for both saccade tasks, F(1, 14) 7.9, P < 0.05 for the prosaccades and F(1, 14) 25.8 with P < 0.001 for the endogenous saccades. 5.3. Discussion This experiment shows that interference is not restricted to saccade tasks in which a prepotent response must be inhibited. In the endogenous saccade task participants
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
81
Fig. 3. Latencies in ms for prosaccades (PS) and endogenous saccades (ES) as a function of load in Experiment 3.
could only generate saccades in a controlled way, due to the lack of peripheral stimuli. In the exogenous saccade task correct saccades can be generated automatically, but some participants, or perhaps even all participants on some trials, could have controlled their eye movements. Although a slight interference eect was found on the exogenous prosaccades, the eect of the RIG task was much larger on the endogenously generated saccades. This eect supports the hypothesis that the performing of a prosaccade in a controlled fashion increases the vulnerability of dualtask interference. The manipulation of stimulus presentation (i.e. smaller ®xation point, larger peripheral stimulus) seems to decrease the prosaccade latencies, but apparently cannot prevent that at least some trials (the controlled ones?) were in¯uenced by the RIG load. This seems to suggest that one cannot assume prosaccades to be of a completely automatic nature when studying them in an arti®cial setting like an experiment.
6. General discussion In line with our hypothesis, the ®rst experiment revealed that an executive interference task increased the latencies and the number of errors in the antisaccade task. But, although the antisaccades are more prone to disruption, an eect on the prosaccade latencies was also found. Moreover, a closer inspection of the interaction of saccade task and load showed that the interference eect was equally strong for both the RIG task and the ®xed tapping task in the two saccade tasks. These unexpected ®ndings were addressed in Experiments 2 and 3. Experiment 2 compared the eects of strict and lenient ®xed tapping. It appeared that under more lenient instructions for the ®xed tapping task, there was no eect on either prosaccades or antisaccades. Under more strict instructions, however, there was an eect, but on antisaccades only. As both ®xed tapping tasks require the same
82
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
motor actions, while antisaccades require controlled processing, it seems that the observed interference eect is of a cognitive nature. Interestingly, Experiment 1 has shown that the load of the secondary tasks had an eect on both saccade tasks. The working hypothesis was that prosaccades, in view of their supposed automatic nature, would be insensitive to executive interference. The interference found on the prosaccades therefore is unexpected, if we assume that interference occurs with saccades that need the inhibition of a prepotent response. However, it is possible that the component ``controlled'' is equally or more important in the execution of saccades. Considering the rather high latencies of some of the participants on the prosaccade task in the control condition, it could be that some participants controlled the execution of some of the prosaccades, generating these saccades through the FEF rather than through the CS. Convinced that a planned execution of the prosaccades will increase their performance, they not only increase the latencies on the prosaccade task, but they may also have made this task vulnerable to dual-task interference. The additional analysis revealed that only the latencies of the slower group of participants were in¯uenced by the strict ®xed tapping load. These data only suggest and do not really prove the idea of controlled execution. Presuming this idea is correct, further support for this hypothesis was collected by showing that mere planned execution of a saccade can cause cognitive interference eects. Using an endogenous prosaccade task, Experiment 3 showed that the fact that a saccade is performed in a controlled fashion is indeed a source of interference. Moreover, the amount of interference is comparable to the interference found on the antisaccade task. So that at least in these experiments, no evidence was found for interference due to inhibition of a prepotent saccade over and above the interference eect found on planned saccades. In the experiment of Roberts et al. (1994) with mental arithmetic as secondary task, no interference on the prosaccades was found. A possible explanation is based on the fact that mental arithmetic loads more heavily on the executive functions than our tapping tasks. As a result more executive control is needed for this secondary task and less executive capacity is left for the prosaccade task which is then performed with less control, resulting in more re¯exive prosaccades. Support for this idea can be found in an experiment of Crevits et al. (1998). Here, prefrontal patients (excluding FEF lesions) executed the prosaccade and the antisaccade task both as a single task and in combination with the RIG task. Compared to a control group of healthy participants, only the antisaccades were impaired, but not the prosaccades. The latter were not in¯uenced by the RIG task. Here appears a similarity to the heavy load of mental arithmetic, the frontal lesion leaves the participants little capacity for controlled execution of the prosaccades, and consequently these saccades are more likely to be performed in the most natural way, i.e. automatically. A study by Henik, Rafal and Rhodes (1994) reports converging evidence for the importance of the controlled generation of saccades and executive interference: patients with frontal lesions performed worse on an endogenous saccade task, similar to the one we used in Experiment 3. Pashler, Carrier and Homan (1993) also found dual-task interference on an endogenous prosaccade task using a
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
83
psychological refractory period (PRP) paradigm. However, only the latencies of the saccade tasks that had to be executed in a controlled manner, showed a PRP eect, suggesting that in these experiments, the load of the PRP trials was so high that participants generated the prosaccades without control whenever possible. For both the saccade tasks and the secondary tapping tasks, the original focus was on inhibition (of a prepotent saccade, of regular time interval sequences, i.e. the random instruction) as a source of interference. In sum, our results suggest that inhibition here had no additional importance, and the controlled processing of both saccades and tapping behaviour could be the origin of the observed cognitive interference. This can explain the equal eects of RIG and strict ®xed tapping and the eect on the exogenous prosaccades. This eect is of course smaller than the one observed with antisaccades and endogenous saccades, because in the latter tasks all saccades have to be planned. In the exogenous prosaccade task on the contrary, only a proportion of the trials could have been planned, and this proportion varied across subjects; in the remaining trials the saccades could be executed automatically. By using secondary tasks that do not rely so heavily on executive control as the mental arithmetic task used by Roberts et al. (1994), it appears that healthy participants under secondary task load are not completely comparable with, for example, prefrontal patients. One methodological note, and a suggestion for further research, is in place here. We believe that the dual-task paradigm is perhaps a research technique at too macroscopic a level. Saccades are known to be carried out rather quickly: depending on circumstances and task requirements, saccades vary between 80 and 450 ms. Tapping behaviour was instructed to be at a pace of about one tap per second. On some trials, the decision to tap could have occurred 800 ms before the programming of the eye movement, on other trials it could almost have coincided with it. For this reason, we believe that designs which allow a trial per trial analysis of the eect of temporal overlap could oer a more microscopic and ecient methodology to investigate the interaction of eye movement control and other executive functions. One possibility of this trial by trial method is to use response-response compatibility as a factor aecting saccade performance in dual-task conditions. Currently, we use this approach to investigate dierences in action-planning for automatic and controlled saccades.
Acknowledgements The authors are indebted to Marc Brysbaert, Gino De Vooght, Gezinus Wolters and two anonymous reviewers for their valuable comments on an earlier version of this paper and to Eva Kemps for checking and improving the English. Part of the research reported in this paper was supported by grant G0012.97 from the Fund for Scienti®c Research-Flanders to Andre Vandierendonck and a doctoral scholarship from the Research Fund of the University of Gent to Els Stuyven (BOF no. 011D0896).
84
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
References Baddeley, A. D. (1966). The capacity for generating information by randomisation. The Quarterly Journal of Experimental Psychology, 18, 119±129. Baddeley, A. (1986). Working memory. Oxford University Press: Oxford. Baddeley, A. (1990). Human memory: theory and practice. London: Erlbaum. Baddeley, A. D., & Hitch, J. G. (1974). Working memory. In G. H. Bower, The psychology of learning and motivation, vol. 8. (pp. 47±89). New York: Academic Press. Baddeley, A. D., Lewis, V., Eldridge, M., & Thomson, N. (1984). Attention and retrieval from long-term memory. Journal of Experimental Psychology (General), 113, 518±540. Claeys, K., Crevits, L., Stuyven, E., Van der Goten, K., Depuydt, C., & Vandierendonck, A. (1999). Parallel visual and memory processes. Documenta Opthalmologica, 95, 349±358. Crevits, L., & De Ridder, K. (1997). Disturbed striato-prefrontal mediated visual behaviour in moderate to severe parkinsonian patients. Journal of Neurology Neurosurgery Psychiatry, 63, 296±299. Crevits, L., Claeys, K., Stuyven, E., Depuydt, C., Van der Goten, K., & Vandierendonck, A. (submitted). Saccadic behaviour and working memory in prefrontal patients, submitted. Dalrymple-Alford, J. C., Kalders, A. S., Jones, R. D., & Watson, R. W. (1994). A central executive de®cit in patients with Parkinson's disease. Journal of Neurology Neurosurgery and Psychiatry, 57, 360± 367. De Rammelaere, S., Stuyven, E., & Vandierendonck, A. (1999). The contribution of working memory resources in the veri®cation of simple mental arithmetic sums. Psychological Research, 62, 72±77. Fukushima, J., Fukushima, K., Chiba, T., Tanaka, S., Yamashita, I., & Kato, M. (1988). Disturbances of voluntary control of saccadic eye movements in schizophrenic patients. Biological Psychiatry, 23, 670± 677. Fukushima, J., Fukushima, K., Miysaka, K., & Yamashita, I. (1994). Voluntary control of saccadic eye movements in patients with frontal cortical lesions and parkinsonian patients in comparison with that in schizophrenics. Biological Psychiatry, 36, 21±30. Fukushima, J., Fukushima, K., Morita, N., & Yamashita, I. (1990). Further analysis of the control of voluntary saccadic eye movements in schizophrenic patients. Biological Psychiatry, 28, 943±958. Gathercole, S. E. (1994). Neuropsychology and working memory: a review. Neuropsychology, 8, 494±505. Gilhooly, K. J., Logie, R. H., Wetherick, N. E., & Wynn, V. (1993). Working memory and strategies in syllogistic-reasoning tasks. Memory and Cognition, 21, 115±124. Glanzer, M., Dorfman, D., & Kaplan, B. (1981). Short-term storage in the processing of text. Journal of Verbal Learning and Verbal Behavior, 20, 656±670. Greenwald, A. G., & Shulman, H. G. (1973). On doing two things at once: II Elimination of the psychological refractory period eect. Journal of Experimental Psychology, 101, 70±76. Guitton, D., Buchtel, H. A., & Douglas, R. M. (1985). Frontal lobe lesions in man cause diculties in suppressing re¯exive glances and in generating goal-directed saccades. Experimental Brain Research, 58, 455±472. Hallet, P. E., & Adams, B. D. (1980). The predictability of saccadic latency in a novel voluntary oculomotor task. Vision Research, 20, 329±339. Hallet, P. E. (1978). Primary and secondary saccades to goals de®ned by instructions. Vision Research, 18, 1279±1296. Henik, A., Rafal, R., & Rhodes, D. (1994). Endogenously generated and visually guided saccades after lesions of the human frontal eye ®elds. Journal of Cognitive Neuroscience, 6, 400±411. Kitagawa, M., Fukushima, J., & Tashiro, K. (1994). Relationship between antisaccades and the clinical symptoms in Parkinson's disease. Neurology, 44, 2285±2289. Logie, R. H., Gilhooly, K. J., & Wynn, V. (1994). Counting on working memory in arithmetic problem solving. Memory and Cognition, 22, 395±410. Martein, R., Kemps, E., & Vandierendonck, A. (1999). The role of working memory in a double span task. Psychologica Belgica, 39, 15±27. McCall, R. B., & Appelbaum, M. I. (1973). Bias in the analysis of repeated measure designs: some alternative approaches. Child Development, 44, 401±415.
E. Stuyven et al. / Acta Psychologica 104 (2000) 69±85
85
Pashler, H., Carrier, M., & Homan, J. (1993). Saccadic eye movements and dual-task interference. The Quarterly Journal of Experimental Psychology, 46A, 51±82. Rapee, R. M. (1993). The utilisation of working memory by worry. Behaviour Research Therapy, 31, 617±620. Roberts, R. J., Hager, L. D., & Heron, C. (1994). Prefrontal cognitive processes: working memory and inhibition in the antisaccade task. Journal of Experimental Psychology: General, 123, 374±393. Rosse, R. B., Schwartz, B. L., Kim, S. Y., & Deutsch, J. I. (1993). Correlation between antisaccade and wisconsin card sorting test performance in schizofrenia. American Journal of Psychiatry, 150, 333±335. Rusted, J. M. (1988). Dissociative eect scopolamine on working memory in healthy young volunteers. Psychopharmacology, 96, 487±492. Stuyven, E., & Van der Goten, K. (1995). Stimulus independent thoughts and working memory: the role of the central executive. Psychologica Belgica, 35, 241±251. Vandierendonck, A., De Vooght, G., & Van der Goten, K. (1998). Does random time interval generation interfere with working memory executive functions? European Journal of Cognitive Psychology, 10, 413±442.