The influence of time pressure and cue validity on ... - Research

induced arousal (like in Posner's, 1980, concept of automatic alertness) and ... evaluation of both the time needed for response initiation and their progress.
155KB taille 19 téléchargements 399 vues
Acta Psychologica 105 (2000) 89±105

www.elsevier.com/locate/actpsy

The in¯uence of time pressure and cue validity on response force in an S1-S2 paradigm Piotr Jaskowski b c

a,b,*

, Rob H.J. van der Lubbe a, Bernd Wauschkuhn a, Edmund Wascher c, Rolf Verleger a

a Department of Neurology, Medical University of L ubeck, 23538 Lubeck, Germany Department of Psychophysiology, Casimirus The Great University of Bydgoszcz, Poland Department of Clinical and Physiological Psychology, University of T ubingen, Germany

Received 27 August 1999; received in revised form 4 May 2000; accepted 24 May 2000

Abstract Hypotheses about variations of response force have emphasised the in¯uences of arousal and of motor preparation. To study both types of in¯uences in one experiment, the e€ects of time pressure and of validity of S1 were investigated in tasks wherein a ®rst stimulus (S1) indicated the most probable response (80% valid) required after a second stimulus (S2). Under time pressure, responses were executed more forcefully while, as could be expected, response times were shorter and errors were more frequent. This pattern of results was not only obtained when time pressure was varied between blocks, but also when varied from trial to trial, by information given by S2. Also invalidly cued responses were executed more forcefully but, as could be expected, in contrast to time pressure, response times were longer and errors were more frequent. The results demonstrate that latency and force of responses may vary in different directions. Ways are outlined on how current hypotheses must be extended in order to account for these results. Ó 2000 Elsevier Science B.V. All rights reserved. Keywords: Response force; S1-S2 paradigm; Time pressure; Cue validity

*

Corresponding author. Tel.: +49-451-500-3706; fax: +49-451-500-2489. E-mail address: [email protected] (P. JasÂkowski).

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 ( 0 0 ) 0 0 0 4 6 - 9

90

P. Jaskowski et al. / Acta Psychologica 105 (2000) 89±105

1. Introduction Recently, interest has been revived in response force as a parameter to study motor response generation in humans (e.g., Angel, 1973; Abrams & Balota, 1991; Jaskowski & Verleger, 1993; Ulrich & Mattes, 1996; Miller, Franz & Ulrich, 1999). Of special interest in these studies was the peak amplitude of the force course, usually referred to as peak force, or simply as response force. Being a parameter of the intensity of motor execution, response force is generally assumed to re¯ect more directly the ®nal than the early stages of the processing chain from stimulus to response, and is therefore expected to depend on variables which are known to a€ect motor processes. For instance, response force was shown to depend on foreperiod duration, being largest after shortest foreperiods (400 ms, Giray, 1990; Ulrich & Mattes, 1996). However, contrary to that simple assumption, response force is also a€ected by some variables which in¯uence the early stages of information processing, e.g. stimulus intensity (Angel, 1973; Jaskowski, Rybarczyk, Jaroszyk & Lema nski, 1995; Miller et al., 1999; Ulrich, Rinkenauer & Miller, 1998). The ®rst theoretical conception on how such ``early'' e€ects might arise was proposed by Giray (1990); see also Giray and Ulrich (1993). In that study, response force proved to depend on the duration between warning signals and imperative stimuli, being larger after shorter foreperiods. Giray (1990) suggested that this e€ect was due to the transient arousing e€ects of the warning stimuli. According to this hypothesis, which we will call ÔarousalÕ hypothesis, the main chain of processing stages starting with stimulus preprocessing and ending with motor execution is paralleled by a bypass from preprocessing to motor execution, consisting of an energetical channel (cf. Sanders' model of stress, 1983) that is activated by stimulusinduced arousal (like in Posner's, 1980, concept of automatic alertness) and directly a€ects response dynamics. This hypothesis can explain why the intensity of auditory stimuli a€ects response force and why no such e€ect is observed with visual stimuli (Jaskowski et al., 1995; see however Ulrich and Mattes, 1996; Ulrich et al., 1998) since visual stimuli are considered to be less arousing than auditory stimuli when their retinal size is small and their intensity is below 650 cd/m2 (Niemi & Lehtonen, 1982; Sanders, 1975). Next to these transient e€ects of stimuli (which may be denoted as immediate arousal, e.g. Sanders, 1983), this hypothesis may also be used to explain di€erences between conditions. For instance, Jaskowski, Wr oblewski and Hojan-Jezierska (1994b) found larger force in conditions wherein participants occasionally received task-irrelevant electrical shocks. They explained these results by applying GirayÕs hypothesis to di€erences in general arousal between conditions rather than immediate arousal evoked by single stimuli. In addition, the increase of response force induced by between-blocks variation of time pressure (Jaskowski, Verleger & Wascher, 1994a) and by knowledge of results (Jaskowski & Wµodarczyk, 1997) can be explained by this generalized arousal hypothesis. Since GirayÕs work, the number of hypotheses has proliferated almost as fast as the number of the experiments concerning response force (Jaskowski & Verleger, 1993; Ulrich & Mattes, 1996; Mattes, Ulrich & Miller, 1997; Ulrich et al., 1998). Unfortunately, no hypothesis can explain all the results gathered so far. For

P. Jaskowski et al. / Acta Psychologica 105 (2000) 89±105

91

instance, the arousal hypothesis fails to explain the e€ects of expectancy. Jaskowski and Verleger (1993) manipulated expectancy for the imperative stimulus without using warning signals, by presenting imperative stimuli with higher probability when a rotating clock-like pointer was at 12 h. Because there are no warning stimuli which could arouse the participants, GirayÕs original arousal hypothesis predicts no changes of response force and Jaskowski et al.Õs (1994b) generalized arousal hypothesis would, if anything, predict decreases of response force in time epochs when stimuli were less probable to occur, due to a reduced level of general activation. However, response force was larger when stimuli had a lower probability of occurrence. These results were recently con®rmed by Mattes and Ulrich (1997) and extended by Mattes et al. (1997) to situations where responses rather than stimuli had a lower probability of occurrence. To account for their results, Jaskowski and Verleger (1993) suggested (see Jaskowski et al., 1994a; Jaskowski and Wµodarczyk, 1997, for more explicit formulations) that at the moment of stimulus presentation, participants make an immediate evaluation of both the time needed for response initiation and their progress of preparation. The result of this evaluation could Ôon-lineÕ modulate arousal/activation according to task demands. Thus, participants would attempt to compensate for poor preparation by increasing arousal/activation which, in turn, would enlarge response force. We will refer to this conjecture as the ÔcompensationÕ hypothesis. Thus, the combination of GirayÕs arousal hypothesis, and Jaskowski and VerlegerÕs compensation hypothesis of arousal provides an explanation for a number of ®ndings on response force. However, neither Giray nor Jaskowski and Verleger attempted to explain the precise mechanism by which the increase of arousal/activation causes a change of force. Such a mechanism was recently proposed by Van Galen and De Jong (1995), based on their investigations of axial pressure exerted on a pen during aiming movements. Van Galen and coworkers (Van Gemmert & Van Galen, 1997; Van Gemmert & Van Galen, 1998; Van den Heuvel, Van Galen, Teulings, Van Gemmert, 1998) found that axial pen pressure under stress evoked by mental load, physical stress or task demands is greater than in a relaxed state. Similar to the arousal hypothesis, their model assumes that sensory stimulation, warning signals and e€ort mechanisms induce stress. Additionally, stress is assumed to increase the level of neuromotor noise, which, on the one hand, ``has an activating and alerting function on the processing capacities of the system as a whole'' (Van Gemmert & Van Galen, 1997, p. 1300) but, on the other hand, leads to a smaller signal-to-noise ratio in the system. To optimize signal-to-noise ratio, a biomechanical strategy is applied: the increased noise is damped and ®ltered by increased sti€ness of the active limbs, i.e., by co-contracting both the agonist and the antagonist muscles, leading to increased axial pen pressure. Although the model was originally developed for aiming movements, it can be applied for ballistic movements. Limb sti€ness can account for a number of results obtained for response force, especially for the e€ect of stress factors like time pressure or task-irrelevant electric shocks. In line with Jaskowski and VerlegerÕs hypothesis (1993), limb sti€ness is a strategy developed to cope with task demands. The neuromotor hypothesis by Van Galen and De Jong (1995), Van Gemmert and Van

92

P. Jaskowski et al. / Acta Psychologica 105 (2000) 89±105

Galen (1997) postulates a kind of compensation mechanism but, unlike Jaskowski and Verleger, this compensation mechanism is designed to reduce neuromotor noise which otherwise might produce unwelcome e€ects like tremor or other signs of motor restlessness. Moreover, recent ®ndings (Van Den Heuvel et al., 1998) indicate that an abrupt change of task demands may evoke larger axial pen pressure, thus that the process of muscle adaptation leading to sti€ness is very ¯exible, in line with Jaskowski and Verleger's suggestion. An alternative approach to account for e€ects on response force was taken by Mattes et al. (1997). They found that response force was larger when the overall probability for making a response was small (in go/no-go tasks). They explained this result with reference to N a at anenÕs (1971) and Niemi and Naatanen (1981) idea of motor readiness. In N a at anenÕs original idea, motor readiness is determined by an interplay between motor excitatory and inhibitory commands. Its level is controlled centrally to maintain an optimal level of motor preparation. The overt response is executed if motor readiness crosses a threshold level, called motor-action limit. The more advanced motor preparation is, the smaller is the distance between the current value of readiness and the motor-action limit. When the stimulus appears, more excitatory commands are delivered and readiness crosses the motor-action limit. According to this model, reaction time (RT) depends on the distance between readiness and motor-action limit, being shorter when this distance is smaller. Mattes et al. (1997) tested some additional assumptions of NaatanenÕs model to account for the e€ect of response probability on response force. According to the version of the model that was supported by their data, the overshoot model, response force is related to the absolute level of motor activation. It was assumed that a large distance between motor readiness and the motor-action limit needs a larger increment of motor activation to exceed the threshold. This leads, in turn, to an overshoot and thereby to more forceful responses. Thus, this model can account for small force and short RT of well-prepared responses as well as for strong force and long RTs of unprepared responses. Like Giray (1990) as well as Jaskowski and Verleger (1993), Mattes et al. (1997) did not try to complete their model with biomechanical considerations. The present set of experiments was designed to test the four hypotheses described so far in an S1-S2 paradigm in which preparation was manipulated by using response cues (S1) indicating the most probable response required after a second stimulus (S2), and by varying time pressure either between blocks (Experiments 1 and 2), or also on a trial-by-trial basis by S2 (Experiment 3). The four models diverge in their predictions about the resulting behavior of response force. Thus, the experiments served both to extend the data base for models of response force and to test the hypotheses proposed so far. 2. Experiment 1 A distinction can be made between two groups of factors (independent variables) with regard to their e€ects on RT, on response correctness, and on response force.

P. Jaskowski et al. / Acta Psychologica 105 (2000) 89±105

93

On the one hand, some manipulations make the response both faster, less accurate, and stronger. These are, for example, loudness (Angel, 1973; Jaskowski et al., 1995), brightness (Angel, 1973; Ulrich et al., 1999; but see Jaskowski et al., 1995), time pressure (Jaskowski et al., 1994a) and knowledge of results (Jaskowski & Wµodarczyk, 1997). This group of factors may be denoted as (either immediately or generally) arousing or activating factors, and they may be characterized as inducing a tradeo€ between speed and accuracy (e.g. see Pachella, 1974). On the other hand, other variables make the response faster, more accurate, and response force weaker, for example, expectancy (Jaskowski & Verleger, 1993; Mattes & Ulrich, 1997) and increased response probability (Mattes et al., 1997). This group of factors may be denoted as motor preparation factors (either speci®c or unspeci®c). In the ®rst experiment, we manipulated two variables which were thought to represent factors of both groups: time pressure, a factor of the ®rst group, and validity with which a cue (the S1) announced the response required after the imperative stimulus (S2), a factor of the second group. The arousal hypothesis can account for the e€ects of the variables from the ®rst group (larger forces with shorter RTs and decreased accuracy) but cannot explain the e€ects of the second group (larger forces with longer RTs) without making additional assumptions about the time course of immediate arousal evoked by the onset of S2. These assumptions are provided by the compensation hypothesis: If, after identi®cation of S2, participants can on-line compensate for the inappropriate preparation induced by invalid S1 information, then response force should increase although RTs are slower. This is in contrast to the original arousal hypothesis which can only predict that response force should decrease when RTs are slower. According to Van Gemmert and Van Galen (1997), stress has an overall activating e€ect on the organism, which leads to an increase of neuromotor noise. The changes of sti€ness, which develop to reduce the consequences of neuromotor noise, are assumed to be ``a strategic adaptation'' (p. 1305) but are also assumed to be fast enough to adapt to abrupt changes of the experimental situation. Therefore, this hypothesis not only predicts a strong e€ect of time pressure on response force, but may also predict an increase of response force after an invalid S1 due to increased muscle sti€ness, in line with the compensation hypothesis. The overshoot hypothesis can account for the variables of the second group when preparation (either speci®c or unspeci®c) is poor. That is, response force should increase in case of poor preparation, because larger overshoots are expected under such conditions due to the greater distance between the initial readiness level and the motor limit. Therefore, after valid cues, the correct response has been prepared, which implies a short distance between motor readiness and the motor limit, whereas after invalid cues, the distance between motor readiness and the motor limit will be relatively large for the required response. Thus, response force should be larger and responses should be slower when imperative signals are invalidly cued due to a greater distance between the readiness level and the motor limit. With regard to e€ects of time pressure, the authors of the hypothesis (Mattes et al., 1997) unfortunately gave no hints as to whether or not other factors could a€ect the distance between the readiness level and the motor limit, or possibly the rate of motor

94

P. Jaskowski et al. / Acta Psychologica 105 (2000) 89±105

activation. However, if we assume that the overshoot hypothesis inherits all features of the original motor-limit hypothesis, then we can expect that due to time pressure the responses should be weaker, because Niemi and Naatanen (1981, p. 158), stated that if speed is stressed in the instructions then the average distance between readiness level and motor limit is smaller. To summarize, of the four hypotheses, the arousal hypothesis in its generalized version predicts a decrease of response force after invalid cues whereas the other three hypotheses predict an increase. For the e€ects of time pressure, the overshoot hypothesis would tend to predict a decrease of response force, whereas the other three hypotheses predict an increase. 2.1. Method 2.1.1. Participants Fourteen (eight males and six females) participants, aged 22±29 yrs, recruited from a population of medical students, took part in the experiment. All participants were right-handed, in good physical health, and had no history of psychiatric or neurological disorders. 2.1.2. Stimuli and procedure The ®rst stimulus (S1) was a white arrowhead (1.2 ´ 1.2 cm2 ), placed in the center of two light-gray concentric circles in the center of an otherwise black screen. The outer circle was un®lled and had a diameter of 6 cm (visual angle ˆ 3°), the inner circle was ®lled in light-gray and had a diameter of 3.7 cm (1.8°) in the condition with high time pressure and of 3 cm (1.5°) in the condition with low time pressure. The arrow pointed to the left or right indicating the most probable response required after S2 with a validity of 80%. After one second, the color of the inner circle changed from gray to blue, forming the imperative stimulus S2. On invalidly cued trials, the arrow changed direction. Simultaneously, the outer circle began ®lling inwards towards the inner circle. The participantsÕ task was to respond with the correct hand before the space was completely ®lled. The screen was cleared 2 s after S2 onset and a message was presented for 600 ms that informed participants whether their response was correct and fast enough. The next trial started 1.1 s after feedback onset, thus the interval between two S1 onsets was 4.1 s. To keep time pressure on an individually equal level, the ®lling speed was changed by an adaptive ``staircase'' method (e.g., Jaskowski et al., 1994a) in the condition with high time pressure, separately for each type of response (left and right, validly and invalidly cued): after two consecutive correct and fast-enough responses the ®lling time was decreased, making the task more dicult; after an incorrect or too slow response ®lling the time was increased, making the task easier. The steps for decreasing and increasing were 60 ms at the beginning of blocks, to reach threshold rapidly, but were then reduced to 15 ms. In the condition with low time pressure, the ®lling speed was kept constant and participants had enough time (600 ms) to make their response. The starting value of the ®lling time for the condition with high time pressure was 450 ms. Participants were told that ®lling speed varied randomly in the

P. Jaskowski et al. / Acta Psychologica 105 (2000) 89±105

95

condition with high time pressure. The order of conditions was counterbalanced across participants. In either condition, 400 stimuli were presented, 320 (80%) trials with valid information provided by S1, and 80 (20%) trials with invalid information. 2.1.3. Apparatus Participants were seated in a comfortable armchair in a sound-proof chamber and viewed a Multisync color monitor from a distance of approximately 130 cm. Stimulus presentation at this monitor was controlled by a PC. Participants were asked to ®x their gaze on the ®xation cross that was visible in the center of the screen. Participants responded with their index ®ngers by pressing one of two response keys equipped with a mechano-to-electrical converter (force sensing elements dismounted from an electronic scale) which was fed to the computer via a 12-bit A/D system. The keys did not bend under the exerted force. In Experiments 1 and 3, a short tone was provided when the force output exceeded 2 N, informing the participants that their response was registered. In Experiment 2 no such feedback was employed. 2.1.4. Data analysis and statistical evaluation RT was de®ned as the time interval between stimulus onset and the moment when force output reached a criterion of 2 N. This value is well in the range used for all-ornone response keys that are normally used in RT experiments. Response force was de®ned as the maximum value of force output. Both RT and response force were determined from the force-time curves separately for each trial. Trials with premature (