Psychological Research (2000) 64: 93±104

Ó Springer-Verlag 2000

ORIGINAL ARTICLE

Marielle Wienese á Wido La Heij A. H. C. van der Heijden á Richard M. Shi€rin

Perceptual inertia: spatial attention and warning foreperiod?

Received: 27 September 1999 / Accepted: 11 February 2000

Abstract Perceptual inertia refers to a potential explanation for the observation that during a short period after stimulus onset, the visual system is insensitive to discriminatory detail. The present study attempted to replicate this empirical result in a simple one-item letteridenti®cation task. The results provided little if any support for the construct of perceptual inertia. In the ®ve experiments reported, evidence consistent with perceptual inertia was only obtained in Experiment 5. However, the Experiment 5 results can at least equally parsimoniously be explained in terms of two standard explanatory factors: an e€ect of precuing of the target position (selective attention) and foreperiod duration (general preparation).

Perceptual inertia Neisser (1967, pp. 16±17) states that, if we are to understand human visual information processing, we must begin by abandoning a set of ``naive realist'' assumptions. One of the assumptions that has to be abandoned is the assumption that a participants' visual experience begins when a pattern is ®rst exposed and terminates when it is turned o€. Within information processing psychology iconic memory research provided abundant evidence in favor of the assumption that information processing does not stop at stimulus termination (e.g., Sperling, 1960; see Coltheart, 1980, for

M. Wienese á W. La Heij (&) á A. H. C. van der Heijden Leiden University, Faculty of Social Sciences, Department of Experimental and Theoretical Psychology, P.O. Box 9555, 2300 RB Leiden, The Netherlands; e-mail: [email protected] R. M. Shi€rin Indiana University, Bloomington, Indiana, USA

an overview). Some evidence in favor of the assumption that visual information processing does not begin at stimulus presentation, but that there is some initial ``dead time'', was provided by Ho€man (1975), Eriksen, Webb, and Fournier (1990) and Shi€rin, Diller, and Cohen (1996). This evidence, suggesting an initial period with ``perceptual inertia'' (Eriksen et al., 1990), is the topic of the present study. To ease exposition, in what follows we ®rst characterize in general terms the main evidence. Then we describe the details of the relevant research. Although the studies di€er considerably in approach, the relevant phenomenon is quite simple (see Fig. 1). Imagine a (n choice) response time (RT) task with two conditions. In both conditions the response latency is measured from the moment that the imperative target, T, appears. In the T-T condition the target appears a ms after the trial begins (in the interval, a, between trial begin and target presentation the target position is empty). In this condition a mean RT of RT(a) is observed. In the N-T condition, after the empty interval of a ms, ®rst irrelevant information (the prime, N) is presented for x ms. The target T appears a + x ˆ b ms after the onset of the trial. In this condition a mean RT of RT(b) is observed. The phenomenon of interest is the fact that RT(a) and RT(b) are not approximately equal but that RT(a) is d ms larger than RT(b) (with d substantially larger than 0). This result can be taken to indicate that information processing starts with an interval of d ms of ``general operations'' in which detailed visual information is not used and only approximate information is required. In this view the irrelevant information, N, is used to start the processing without harm to the eventual response because the system is not yet processing the details of the stimulus. Thus, in the T-T condition the ``general operations'' take place after target onset [so d is included in RT(a)], but in the N-T condition the general operations take place before target onset [so d is not included in RT(b)]. Therefore, RT(a) is d ms larger than RT(b). The interval, d, of ``un-used'' time in RT(a) is referred to as dead time (Ho€man,

94

Fig. 1 A schematic representation of the two main presentation conditions in experiments on perceptual inertia (N moment of presentation of the irrelevant information, T moment of presentation of the target, d period of ``dead time'', RT response time). For further explanation, see text

1975). It is accounted for in terms of perceptual inertia (Eriksen et al., 1990)1. In this article we use the term ``perceptual inertia'' to refer both to the empirical result and the theoretical hypothesis explaining the result; it will be clear from the context which is meant. We now turn to the details of the relevant research. Ho€man (1975) sought to determine the nature and temporal course of the interference provided by nonattended visual material. Participants' RTs were measured for deciding which of two targets, a 3 or a 5, occupied an indicated position in eight-element displays and one-element displays. Ho€man's main evidence for a dead time comes from his ``target-delayed'' condition. In both relevant trial types, the target position was indicated by a bar at time t ˆ a. In condition S(imultaneous) the target and the noise items were presented simultaneously at this time. In condition D(elayed) the noise items were presented at this time along with a non-indicative subset of the features of the target. The remaining features of the target that made target identi®cation possible were added 50 ms later, at t ˆ a + 50. Ho€man measured RTs from the time of presentation of the complete target, i.e, from t ˆ a with trial type S and from t ˆ a + 50 with trial type D, and found that the mean RT for trial type S was about 50 ms larger than the mean RT for trial type D.

1 Strictly speaking there are probably two components to the construct of perceptual inertia: The weaker view is the hypothesis that some sorts of information are extracted sooner than others, when the target stimulus starts and does not change; the stronger view is that there is a period of time after onset during which the target stimulus can be replaced by quite di€erent information, without a€ecting the course of processing of the target.

For Ho€man (1975, p. 352) this outcome ± 50 ms simply wasted ± indicated that the visual information processing system, presented with an array of characters, is indi€erent to the presence of the de®ning features of those characters for approximately 50 ms. He states: ``...by starting the clock [in condition D] 50 ms after display onset, we are subtracting 50 ms dead time (insofar as the presence of a complete target is concerned). Thus it appears that the visual information processing system does not use the complete target or noise element information for the ®rst 50 ms when confronted with a multicharacter display.'' (p. 352 [insert ours]). Ho€man further refers to Eriksen and Eriksen (1974) and Rohrbaugh and Eriksen (1975) who reported that RT is indi€erent for short time periods to the presence of complete information for individual letters in a word and for separate components of a visual pattern (Ho€man, 1975, p. 352). In the, for present purposes, most relevant experiment and conditions of the study of Eriksen et al. (1990), attention was directed to 1 of 8 locations in the visual ®eld through an underline precue. A circular array of letters followed the cue after 50 ms. An additional 50 ms after the letter array began, a second underline cue appeared in another position, a position that until the cue appeared contained one of two possible items: a target, T, (a H or an N) or a noise letter, N (another letter). Simultaneously with the occurrence of the second cue, the letter in that cued location either continued unchanged (T-T condition), changed from an irrelevant noise letter to one of the targets (N-T condition), or changed from a target to the other target (T2-T condition). The whole display was turned o€ 50 ms after appearance of the second cue. The subjects had to indicate as fast as possible the identity of the target. RTs were measured from the moment of presentation of the initial circular array of letters. For Eriksen et al. the important ®nding was that the mean RTs in the three conditions did not di€er reliably or appreciably (there was no facilitation in the T-T condition and no interference in the T2-T condition, compared with the N-T condition). Because the identity of the information in the target location in the ®rst 50 ms did not a€ect RT, Eriksen et al. (1990) concluded that the ®rst 50 ms of stimulation is dead time as far as processing of information is concerned. In their view, in this period the presence of discriminatory detail cannot be used. ``There is a phenomenon similar to perceptual inertia during which the processing system is being turned on, so to speak, and until the system is revved up, the presence of discriminatory detail cannot be used.'' (Eriksen et al., 1990, p. 481.) As further evidence for the existence of an initial dead time Eriksen et al. refer to some other experiments in which comparable phenomena were observed (Eriksen & Eriksen, 1974; Rohrbaugh & Eriksen, 1975) and the work reported by Ho€man (1975) that we just described. Shi€rin et al. (1996) used a paradigm related to the one of Eriksen et al. (1990). Subjects had to identify a target, T, (an X or an O) in a circular display of eight

95

characters, otherwise ®lled with noise letters, N (other letters). In the, for present purposes, most relevant experiment and conditions, a cue, presented 185 ms in advance of any letters, directed attention to the location that would contain the target. Then a ®rst display of letters appeared for 150 ms. This prime display might contain an X, an O, or neither, in a random location. Then the target display, containing an X or O in the cued location and noise letters elsewhere, appeared until a response was given. In all conditions RTs were measured from the moment of presentation of the prime display. When the prime occurred in the cued location, RT was speeded up by a compatible prime (T-T condition) and slowed down by an incompatible prime (T2-T condition), both compared with a neutral prime (N-T condition). The crucial observation was, however, that the speeding of T-T relative to N-T was about 75 ms, much less than the 150 ms by which the target in T-T preceded the target in N-T. Shi€rin et al. (1996) ®tted a quantitative model to the data. In this model a parameter d was used. They explain: ``When input ®rst arrives at a location, there is a period of time, d, when general information is accumulated that does not distinguish one character from another, followed by accumulation of distinguishing information. If one input immediately follows another in a location, distinguishing information about the new input begins to be extracted at once, without the imposition of a new period of time, d. The period d corresponds to what C. W. Eriksen et al., (1990) referred to as perceptual inertia. It is needed to account for the fact that the T-T condition is much less than 150 ms faster than the N-T condition.'' (Shi€rin et al., 1996, pp. 238± 239). In a footnote, Shi€rin et al. indicate an alternative interpretation and present additional evidence related to the dead-time hypothesis (we return to this issue in the General discussion). Of course, visual information processing and visual perception are not instantaneous all-or-none processes, but processes that develop gradually and progressively over time. There are many good reasons to believe that upon stimulus presentation there is an initial period in which discriminatory detail is not used or used poorly. This likelihood, however, does not entail that the experiments just mentioned have shown the existence of such a period, or that non-target information presented in this period acts just as target information. Before accepting the construct of perceptual inertia, we decided to take two actions. Firstly, we investigated the reliability and robustness of the basic e€ect. Secondly, in conditions in which the relevant phenomenon can be demonstrated we investigated whether alternative interpretations are possible or preferable. This investigation is critical given the existence of widely explored constructs like ``selective attention'' (the cue may attract attention to the position of the target T, see Fig. 1) and ``general warning e€ects'' (preparation time is longer in the N-T condition than in the T-T condition, see Fig. 1).

Experiment 1 is a simple experiment that, according to the interpretations discussed above, should produce results that can be interpreted as resulting from perceptual inertia. In this experiment the target letter is presented at the point of ®xation and is either preceded by a prime (the N-T condition) or is not preceded by a prime (the T-T condition). To anticipate the results: no evidence of perceptual inertia was obtained. Next, in four experiments we systematically manipulated two factors that might produce e€ects that were previously interpreted as resulting from perceptual inertia: foreknowledge of target position (selective attention) and foreperiod duration (a general warning e€ect). In addition, the results of Experiment 1 prompted the examination of a third factor: ``forward masking''.

Experiment 1 The main purpose of the ®rst experiment was the possibility of demonstrating perceptual inertia in a singlelocation experiment that incorporates and preserves the essential features of the experiments of Shi€rin et al. (1996). In Experiment 1a of Shi€rin et al. a pre-cue, presented 185 ms before letter presentation, directed the participant's attention to one location in visual space. In their N-T condition in the attended position a noise letter, N, appeared for 150 ms (the prime) and then a target, T, until response. In their T-T condition in the attended position a target letter, T, appeared for 150 ms and then continued on the screen until response. For Shi€rin et al. the data relevant to perceptual inertia was the fact that RT in the T-T condition was only about 75 ms faster than RT in the N-T condition (much less than the 150 ms advantage that some models might predict). In our single-location version of the task all letters were presented at the (attended) point of ®xation. The task for the participants was to identify a target letter (O or X) by means of a button-press response. The two main experimental variables were (a) the presence or absence of a (100 ms) prime stimulus (one of the letters E, H or T), and (b) delay (150 versus 250 ms; see further on). Figure 2 shows the four main experimental conditions. Conditions 1 and 2, with a prime stimulus, mimic the N-T condition of Shi€rin et al. (1996) and Conditions 3 and 4, without a prime stimulus, mimic the T-T condition of Shi€rin et al. RTs were measured from moment of target presentation. Evidence for perceptual inertia would be a smaller RT in Condition 1 than in Condition 3, or in Condition 2 than Condition 4. The present experiment and all further experiments varied the time interval, a, between the o€set of the ®xation point and the onset of the target or prime, to investigate the presence of a general warning or preparation e€ect (see, e.g., Posner & Boies, 1971). A general

96 Each participant was given 520 trials. These were divided into 10 series of 52 trials, separated by pauses of approximately 1 min. The ®rst four trials of each series were warm-up trials. The remaining 48 trials consisted of 12 trials of each of the four experimental conditions in a random order. The 24 possible stimuli (2 targets ´ 3 primes ´ 2 delays ´ prime present/absent) were presented two times each. To reduce variance in the data, each incorrect response was followed by a ®ller trial. The results of these ®ller trials were not included in the analyses. At the start of each trial, a ®xation cross was presented at the center of the display for 1,000 ms, followed by a delay of either 150 or 250 ms. In the prime-absent conditions the target (O or X) was then presented until response. In the prime-present conditions a prime (E, H or T) was presented for 100 ms, followed by a target (O or X) until response. The participant was told to ignore the prime, and to react as fast as possible to the target letter while maintaining accuracy. The next trial started 1,500 ms after the participant had responded. RT was measured from the moment of target presentation.

Results Fig. 2 The four main experimental conditions in Experiments 1±5. The e€ect of perceptual inertia is examined by comparing Conditions 1 and 3 and Conditions 2 and 4. The e€ect of preparation time (general warning e€ect) is examined by comparing Conditions 1 and 2 and Conditions 3 and 4

warning e€ect would be evidenced by a smaller RT with a long delay (250 ms) than with a small delay (150 ms): i.e. Condition 2 faster than Condition 1, and Condition 4 faster than Condition 3. Method Participants. Four men and four women, students or employees at the University of Leiden, the Netherlands, served as paid participants. All reported normal or corrected-to-normal vision. Apparatus. The stimuli were presented on a fast display screen (Vector General). RT was determined with an accuracy of 1 ms by means of two push-buttons. Stimulus presentation and registration of RTs and number of errors were controlled by a PDP-11/34 computer. Stimuli. Stimuli were white capital letters on a dark gray background. The letters subtended approximately 0.4° of vertical visual angle. They were presented at the center of the display screen and were viewed binocularly from a distance of 140 cm. Targets were O and X and primes E, H and T. Design. The experiment was designed as a two-factor factorial design with repeated measures on all factors. The ®rst factor was the presence or absence of a prime. The second factor was the size of the delay, 150 or 250 ms. The combination of the two factors is presented in Fig. 2. Procedure. The participants were run individually in a dimly illuminated room. Their task was to discriminate between the target letters O and X as quickly and accurately as possible. Four participants had to push the left button for O and the right button for X and four participants had to push the right button for O and the left button for X.

RTs of erroneous responses were ®rst excluded. Next, for each participant means were calculated for each of the four experimental conditions shown in Fig. 2. Table 1 presents the mean RTs and the percentages of errors of these four conditions. In addition, Table 1 shows (a) the e€ect of the presence of a prime (a positive value indicates that the prime resulted in an increase in RT and error percentage, the opposite of perceptual inertia) and (b) the size of the general warning e€ect (a positive value indicates that an increase in delay resulted in a decrease in RT and error percentage). An analysis of variance (ANOVA) was performed on the mean RTs per participant, with prime (present versus absent), delay (150 ms versus 250 ms) as within-participant factors. In addition, the factor series (1±10) was entered in this analysis and in all analyses to follow in this study. However, because the factor series did not show any interesting results, this factor will not be reported. This analysis showed a signi®cant main e€ect of the factor prime, F(1, 7) ˆ 19.60, p < 0.01.The mean RTs of the prime-present and the prime-absent conditions were 471 ms and 443 ms, respectively. The interaction between prime and delay was signi®cant, F(1, 7) ˆ 19.920, p < 0.01. Inspection of this interaction revealed that an increase in delay facilitated performance in the prime-absent condition, F(1, 7) ˆ 11.605, p < 0.05, but hampered performance in the prime-present condition, F(1, 7) ˆ 5.602, p < 0.05. An identical ANOVA was performed on the error percentages. A signi®cant main e€ect was obtained for the factor prime, F(1, 7) ˆ 5.858, p < 0.05. The presence of a prime resulted in an increase in errors. Discussion A single-location experiment that incorporated and preserved the essential features of Shi€rin et al. (1996) failed to demonstrate perceptual inertia. In fact, the

97 Table 1 Mean response times (in milliseconds) and percentages of errors in the various conditions of experiment 1 RT response time, %e percentage of errors) Prime (N-T)

Delay 150 ms Delay 250 ms General warning e€ect

No prime (T-T)

E€ect prime (N-T)-(T-T)

RT

%e

RT

%e

RT

%e

467 476 )9

3.2 2.0 1.2

451 434 17

2.1 0.9 1.2

16 42

1.1 1.1

results went the opposite way, with Condition 3 (no prime, 150 ms delay) 16 ms faster than Condition 1 (prime, 150 ms delay), and Condition 4 (no prime, 250 ms delay) 42 ms faster than Condition 2 (prime, 250 ms delay). Discussion of possible reasons why Shi€rin et al. obtained di€ering results is deferred to the General discussion. Here we brie¯y look at three standard explanatory factors that could bear on our results. First, selective attention to location (see, e.g., Posner, Snyder, & Davidson, 1980; van der Heijden, 1992). Although the previous studies relevant to perceptual inertia used multiple locations and location cuing, allowing the possibility that this factor played a role (depending on additional assumptions regarding the speed of reallocating spatial attention), the present study presented all stimuli at the attended point of ®xation, and this factor should not have been operative. Second, masking e€ects of a prime (see, e.g., Kahneman, 1968). Conditions 1 and 2 presented a prime and Conditions 3 and 4 did not. It is highly likely that the prime served as a forward mask and hampered target identi®cation. It is possible to argue that perceptual inertia was playing a role in this study, but that the e€ect of masking was larger than that of perceptual inertia. Forward masking was not mentioned by Ho€man (1975), Eriksen et al. (1990) and Shi€rin et al. (1996) as a relevant factor in their experimental paradigms, possibly because their results went in the opposite direction. Third, facilitating e€ects of general preparation (see, e.g., Posner and Boies, 1971, for the e€ects of a warning period). The present experiment showed no main e€ect of delay, but the interaction between prime and delay was highly signi®cant: There was a general warning effect for the no-prime condition, but the opposite e€ect in the prime condition. Further experimental evidence with regard to this factor is desirable. In the following four experiments we systematically varied the three factors mentioned above. This allowed us to determine whether the results obtained require the introduction of the explanatory concept of perceptual inertia or, alternatively, whether these three factors suce to account for the data. One possible important di€erence between our Experiment 1 and the studies by Ho€man (1975), Eriksen et al. (1990), and Shi€rin et al. (1996) was that in the latter studies the position of the target letter varied from

trial to trial. For that reason, in all further experiments the target letter appeared either at the left or at the right of the central ®xation point. In Experiments 2 and 3 the prime appeared both at the target position and at the irrelevant display position, thereby precluding a possible role of selective attention. These two experiments differed in the use of primes that would probably induce a substantial amount of forward masking (a letter prime in Experiment 2) or would lead to no, or only a very small amount of forward masking (a dot prime in Experiment 3). In Experiments 4 and 5, the prime only appeared at the target position, thereby allowing for an e€ect of selective attention. Again, the di€erence between these two experiments was in the type of prime used (a letter prime in Experiment 4 and a dot prime in Experiment 5). As in Experiment 1, in all four experiments the general warning e€ect was examined by manipulating the foreperiod duration.

Experiment 2 There is a clear di€erence between our Experiment 1 and the perceptual inertia experiments earlier reported. In our experiment the prime and target appeared at the point of ®xation, whereas in the experiments reported by Eriksen et al. (1990), Ho€man (1975), and Shi€rin et al. (1996) the target could appear in various positions in the display. It is conceivable that perceptual inertia is only observed under conditions in which an attentional shift towards the target position has to be made. To test this conjecture, in the present experiment locational uncertainty was introduced. In the T-T condition the target appeared unpredictably either to the left or to the right of the point of ®xation. In the N-T condition two identical primes appeared simultaneously before the target, one to the left and one to the right of the point of ®xation. Method Participants. Four men and four women, students or employees at the University of Leiden, the Netherlands, served as paid participants. All reported normal or corrected-to-normal vision. Apparatus. See Experiment 1. Stimuli. Stimuli were white capital letters on a dark gray background. The letters subtended approximately 0.4° of vertical visual angle. They were presented to the left and to the right of the central ®xation point at a distance of 5.6° (center to center). They were viewed binocularly from a distance of 140 cm. Targets were O and X and primes E, H and T. Design. See Experiment 1. Procedure. The procedure was similar to the one used in Experiment 1. The only di€erence was that in each series the 24 di€erent stimuli (2 targets ´ 3 primes ´ 2 delays ´ 2 prime conditions) were

98 presented once to the left and once to the right of the central ®xation point. To ensure that the prime could not be used as a position cue, the prime letter appeared both to the left and to the right of the central ®xation point for 100 ms.

Results The data were treated in the same way as in Experiment 1. Table 2 presents the mean RTs and the percentages of errors of the four main experimental conditions. In addition, Table 2 shows (a) the e€ect of the presence of a prime (a positive value indicates that the prime resulted in an increase in RT and error percentage, the opposite of perceptual inertia) and (b) the size of the general warning e€ect (a positive value indicates that an increase in delay resulted in a decrease in RT and error percentage). An ANOVA was performed on the mean RTs per participant, with prime (present versus absent) and delay (150 ms versus 250 ms) as within-participant factors. This analysis showed a signi®cant main e€ect of the factor prime, F(1, 7) ˆ 42.815, p < 0.00. The mean RTs in the prime condition and the no-prime condition were 503 ms and 476 ms, respectively. The interaction between prime and delay was signi®cant, F(1, 7) ˆ 11.029, p < 0.05. Further inspection of this interaction revealed that an increase in delay facilitated performance in the no-prime condition, F(1, 7) ˆ 12.379, p < 0.05. An identical ANOVA was performed on the error percentages. A signi®cant main e€ect was obtained for the factor prime condition, F(1, 7) ˆ 5.786, p < 0.05. The error percentages in the prime condition and the noprime condition were 6.8% and 2.4%, respectively. Discussion The present experiment introduced location uncertainty, with results that largely duplicated those from Experiment 1. The mean RT in Condition 1 (prime, 150 ms delay) is 17 ms larger than the mean RT in Condition 3 (no prime, 150 ms delay) and the mean RT in Condition 2 (prime, 250 ms delay) is 37 ms larger than the mean RT in Condition 4 (no prime, 250 ms delay), both results in the direction opposite to that predicted by perceptual inertia. Let us consider whether these results can be explained with the standard explanatory factors: selective attention, forward masking and general warning.

Concerning selective attention to location, note that, as in Experiment 1, all conditions were equal with regard to foreknowledge of position. In both the N-T and T-T conditions the target appeared unpredictably either at the left or at the right of ®xation. Thus di€erential attention to location should not have contributed to differences among our conditions. Attention might have been spread more widely across visual space in Experiment 2 than in Experiment 1, but if so this did not change the results. As in Experiment 1, Conditions 1 and 2 (mean RT 503 ms) involved a prime and Conditions 3 and 4 (mean RT 476 ms) did not. It is highly likely that this ®nding results from the prime serving as a forward mask. The ®ndings relevant to the facilitating e€ects of general preparation (due to a warning period before target presentation) are similar to those of Experiment 1: a general warning e€ect for the no-prime condition but not for the prime condition.

Experiment 3 It is highly likely that in Experiments 1 and 2 the letter prime served as a forward mask and hampered target identi®cation. Forward masking may have played a role in the experiments reported by Eriksen et al. (1990) and Shi€rin et al. (1996), but probably not in the experiments reported by Ho€man (1975) in which the prime consisted of a non-indicative subset of the features of the target letter. Because these features are parts of the target and remain when the target is completed, there is every reason to believe that the masking e€ect they produce di€ers from that in e€ect in our experiment. To examine whether forward masking was responsible for the lack of any evidence in favor of perceptual inertia in Experiment 2, Experiment 3 was run. This experiment was very similar to Experiment 2, with the only di€erence that the letter primes used in Experiment 2 were replaced by a small dot. So, in this experiment in the NT conditions two dots appeared, one at the left and one at the right of the ®xation point. After a delay of 100 ms, the two dots disappeared and the target letter was presented either at the left or at the right of the ®xation point. Method

Table 2 Mean response times (in milliseconds) and percentages of errors in the various conditions of experiment 2 Prime (N-T)

Delay 150 ms Delay 250 ms General warning e€ect

No prime (T-T)

E€ect prime (N-T)-(T-T)

RT

%e

RT

%e

RT

%e

499 507 )8

6.2 7.3 )1.1

482 470 12

2.1 2.7 )0.6

17 37

4.1 4.6

Participants. One men and seven women, students or employees at the University of Leiden, the Netherlands, served as paid participants. All reported normal or corrected-to-normal vision. Apparatus. See Experiment 1. Stimuli. See Experiment 2. Instead of the primes E, H and T a small white dot was used that appeared at the center of the position to be occupied by the target letter X or O.

99 Design. See Experiment 2. Procedure. The procedure was similar to the one used in Experiment 2.

Results The data were treated in the same way as in Experiment 1. Table 3 presents the mean RTs and the percentages of errors of the four main experimental conditions. In addition, Table 3 shows (a) the e€ect of the presence of a prime (a positive value indicates that the prime resulted in an increase in RT and error percentage, the opposite of perceptual inertia) and (b) the size of the general warning e€ect (a positive value indicates that an increase in delay resulted in a decrease in RT and error percentage). An ANOVA was performed on the mean RTs per participant, with prime (present versus absent) and delay (150 ms versus 250 ms) as within-participant factors. The main e€ect of the factor prime just failed to reach signi®cance, F(1, 7) ˆ 5.110, p < 0.06. The mean RTs in the prime condition and the no-prime condition were 462 ms and 452 ms, respectively. The interaction between prime and delay was signi®cant, F(1, 7) ˆ 6.644, p < 0.05. Further inspection of this interaction revealed that an increase in delay facilitated performance in the no-prime condition, F(1, 7) ˆ 8.331, p < 0.05. An identical ANOVA was performed on the error percentages. A signi®cant main e€ect was obtained for the factor prime type, F(1, 7) ˆ 18.045, p < 0.01. The error percentages in the prime condition and the noprime condition were 5.7% and 2.3%, respectively. Discussion The results of this experiment support the conjecture that forward masking played a signi®cant role in Experiments 1 and 2. The overall negative e€ect of the presence of a prime stimulus was reduced from 27 ms in Experiment 2 to 10 ms in the present experiment. Nevertheless, there was no indication whatsoever that the use of a prime reduced the response latencies to the target letter (see Table 3). The mean RT in Condition 1 (prime, 150 ms delay) was 3 ms larger than the mean RT in Condition 3 (no prime, 150 ms delay) and the mean RT in Condition 2 (prime, 250 ms delay) was 18 ms Table 3 Mean response times (in milliseconds) and percentages of errors in the various conditions of experiment 3 Prime (N-T)

Delay 150 ms Delay 250 ms General warning e€ect

No prime (T-T)

E€ect prime (N-T)-(T-T)

RT

%e

RT

%e

RT

%e

462 462 0

5.2 6.3 )1.1

459 444 15

2.4 2.3 0.1

3 18

2.8 4.0

larger than the mean RT in Condition 4 (no prime, 250 ms delay), both results in the direction opposite to that predicted by perceptual inertia. Let us consider again the standard explanatory factors: selective attention, forward masking and general warning. Concerning selective attention to location, note that, as in Experiments 1 and 2, all conditions were equal with regard to foreknowledge of position. In both the N-T and T-T conditions the target appeared unpredictably either at the left or at the right of ®xation. Thus, differential attention to location should not have contributed to di€erences among our conditions. As in the previous Experiments, Conditions 1 and 2 involved a prime and Conditions 3 and 4 did not. The results indicate that the forward masking e€ect observed in Experiment 2 was reduced by using a dot prime. The ®ndings relevant to the facilitating e€ects of general preparation (due to a warning period before target presentation) are similar to those of Experiments 1 and 2: a general warning e€ect for the no-prime condition but not for the prime condition.

Experiment 4 Experiments 2 and 3 with position uncertainty showed similar results, opposite to those expected from perceptual inertia. In the experiments of Eriksen et al. (1990) and Shi€rin et al. (1996), knowledge of target position was provided by a precue. Only to the extent that the precue provided enough time to move attention to the target location (much less time for Eriksen et al. than Shi€rin et al.) was foreknowledge of location equated across their relevant conditions. In Ho€man's (1975) experiment, there was de®nitely a 50 ms advantage in foreknowledge in the N-T condition. The purpose of our fourth experiment was to investigate the perceptual inertia issue in a single-item version of the experiment of Ho€man (1975) with foreknowledge of the target position in the N-T conditions and no foreknowledge of the target position in the T-T conditions. The experiment is basically a replication of Experiment 2 with only the N-T conditions modi®ed. In Experiment 2 in the N-T conditions there was no foreknowledge of position because two identical primes were presented, one at the left of ®xation and one at the right of ®xation, before the single target appeared. In the present experiment only one prime was presented either at the left or at the right of ®xation. The target always followed at the position of the prime. So, as in Ho€man's experiment, in the present experiment there was foreknowledge of position in the N-T condition and no foreknowledge of position in the T-T condition. Method Participants. Six men and two women, students or employees at the University of Leiden, the Netherlands, served as paid participants. All reported normal or corrected-to-normal vision.

100 Apparatus. See Experiment 1. Stimuli. See Experiment 2. Design. See Experiment 1. Procedure. The procedure was very similar to the one used in Experiment 2. The only di€erence was that the prime letter was only presented in the position (left or right of the ®xation point) where the target would appear.

Results The data were treated in the same way as in Experiment 1. Table 4 presents the mean RTs and the percentages of errors in the four main experimental conditions. In addition, Table 4 shows (a) the e€ect of the presence of a prime (a positive value indicates that the prime resulted in an increase in RT and error percentage) and (b) the size of the general warning e€ect (a positive value indicates that an increase in delay resulted in a decrease in RT and error percentage). The ANOVA with prime (present versus absent) and delay (150 ms versus 250 ms) as within-participant factors showed a signi®cant main e€ect for the factor delay, F(1, 7) ˆ 9.471, p < 0.05. An identical ANOVA was performed on the percentages of errors. A signi®cant main e€ect was found for the factor prime, F(1, 7) ˆ 10.343, p < 0.05. The error percentages in the prime condition and the noprime condition were 9.6% and 2.7%, respectively. Discussion The main purpose of the present experiment was to investigate the perceptual inertia issue in a version of Ho€man's (1975) experiment with foreknowledge of the target position in the N-T conditions and no foreknowledge of the target position in the T-T conditions. The direction of the relevant di€erences were those predictable from perceptual inertia. However, the differences of 1 and 9 ms were small and far from signi®cant (Ho€man reported a di€erence of 50 ms). Moreover, the percentages of errors are signi®cantly larger in the N-T conditions than in the T-T conditions, consistent with a speed-accuracy trade-o€. Thus, in our Table 4 Mean response times (in milliseconds) and percentages of errors in the various conditions of experiment 4 Prime (N-T)

Delay 150 ms Delay 250 ms General warning e€ect

No prime (T-T)

E€ect prime (N-T)-(T-T)

RT

%e

RT

%e

RT

%e

444 442 2

8.5 10.6 )2.1

453 443 10

2.4 3.0 )0.6

)9 )1

6.1 7.6

Experiment 4 even 100 ms of advance knowledge of location in the N-T condition provided no convincing evidence for perceptual inertia. As in the previous Experiments, Conditions 1 and 2 involved a prime and Conditions 3 and 4 did not. In Experiment 1 and 2 we obtained evidence that the prime served as a forward mask and hampered target identi®cation in the N-T conditions. We, therefore, have to assume that the same forward masking e€ect was operative in the present experiment. The mean RTs in the N-T conditions were, however, slightly smaller than in the T-T conditions. Thus, relative to Experiments 1 and 2, the present situation must have involved a new factor, or a change in the e€ect size of an existing factor, canceling the delaying e€ect of forward masking. As far as possible e€ects of selective attention are concerned it is important to note that, unlike in Experiments 1, 2, and 3, in the present experiment the N-T and T-T conditions di€ered with regard to foreknowledge of position. There was 100 ms foreknowledge of position in the N-T conditions and no foreknowledge of position in the T-T conditions. There is substantial evidence that in the single-item paradigm foreknowledge of position speeds up RTs considerably (see Van der Heijden, 1992, Chapter 4, for a summary of the evidence). It is generally assumed that selective attention is the causal factor involved. It is, therefore, highly likely that in the present experiment selective attention, via foreknowledge of position, considerably facilitated performance in the N-T conditions relative to the T-T conditions and overcame and canceled the delaying e€ect of forward masking in the N-T conditions. As in all previous experiments, in the present experiment we assessed the in¯uence of a general warning e€ect on RT. The results di€ered slightly statistically from those obtained in the Experiments 1, 2 and 3, although the pattern was fairly similar: In the no-prime condition the mean RT in the 250 ms delay condition was 10 ms smaller than in the 150 ms delay condition.

Experiment 5 In Experiment 4, with foreknowledge of position in the N-T conditions and no foreknowledge of position in the T-T conditions as in Ho€man's (1975) experiment, some evidence for perceptual inertia appeared, a slightly faster responding in the N-T conditions than in the T-T conditions (1 and 9 ms). Ho€man, however, reported a much larger e€ect (50 ms). It seems likely that one di€erence between the two tasks can explain this di€erence in magnitude of results. In our task, in the N-T conditions, the letters E, H and T were used as primes. Experiments 1 and 2 indicated that these primes mask the targets O and X. In Ho€man's experiment, in the N-T condition, a (non-indicative) subset of the features of the target served as the prime. Because these

101

features are parts of the target and remain when the target is completed, there is every reason to believe that the masking e€ect they produce di€ers from that in our experiment. The purpose of our last experiment was, therefore, to investigate the perceptual inertia issue in another version of Ho€man's (1975) experiment, with foreknowledge of the target position in the N-T conditions and without foreknowledge of the target position in the T-T conditions, but with forward masking (largely) eliminated. The experiment is basically a replication of Experiment 4 with only the N-T conditions modi®ed. In Experiment 4, in the N-T conditions, the prime that served as the location indicator consisted of one of the letters E, F or H. In the present experiment a small dot, as used in Experiment 3, served as a location indicator. The results of Experiment 3 indicated that the masking induced by a dot prime ± if present ± was substantially smaller than the masking induced by letter primes. Then, as we presume was the case in Ho€man's experiment, in the present experiment there is foreknowledge of position without (much) masking in the N-T conditions and no foreknowledge of position (and no forward masking) in the T-T conditions. Method Participants. Two men and six women, students or employees at the University of Leiden, the Netherlands, served as paid participants. All reported normal or corrected-to-normal vision. Apparatus. See Experiment 1. Stimuli. See Experiment 3. Design. See Experiment 1. Procedure. The procedure was very similar to the one used in our Experiment 4. The only di€erence was that instead of a letter, a dot was used as prime.

Results The data were treated in the same way as in Experiment 1. Table 5 presents the mean RTs and the percentages of errors of the four main experimental conditions. In adTable 5 Mean response times (in milliseconds) and percentages of errors in the various conditions of experiment 5 Prime (N-T)

Delay 150 ms Delay 250 ms General warning e€ect

No prime (T-T)

E€ect prime (N-T)-(T-T)

RT

%e

RT

%e

RT

%e

454 459 )5

3.2 3.1 0.1

500 486 14

1.8 2.2 )0.4

)46 )27

1.4 0.9

dition, Table 5 shows (a) the e€ect of the presence of a prime (a positive value indicates that the prime resulted in an increase in RT and error percentage) and (b) the size of the general warning e€ect (a positive value indicates that an increase in delay resulted in a decrease in RT and error percentage). The ANOVA with prime condition (present versus absent) and delay (150 ms versus 250 ms) as within-participant factors showed a signi®cant main e€ect of the factor prime, F(1, 7) ˆ 135.064, p < 0.001. The mean RTs in the prime-present and prime-absent conditions were 456 ms and 493 ms, respectively. An identical ANOVA was performed on the error percentages. A signi®cant main e€ect was found for the factor prime, F(1, 7) ˆ 6.721, p < 0.05. The error percentages in the prime-present and prime-absent conditions were 3.2% and 2.0%, respectively. Discussion Evidence consistent with the phenomena previously used to identify the presence of perceptual inertia was obtained: The mean RT was 46 ms faster in Condition 1 (prime, 150 ms delay) than in Condition 3 (no prime, 150 ms delay), and 27 ms faster in Condition 2 (prime, 250 ms delay) than in Condition 4 (no prime, 250 ms delay). These di€erences are roughly comparable to those reported by Ho€man (about 50 ms in his study with multiple-character displays and appreciably less in his study with single-character displays). The present experiments (consistent with Ho€man's experiment) show that a smaller RT in the N-T conditions than in the T-T conditions occurs primarily when there is a cue that (a) provides foreknowledge of the target position and (b) does not serve as a forward mask. This outcome suggests an explanation not in terms of perceptual inertia, but in terms of bene®cial e€ects of selective attention in the N-T condition that are not available in the T-T condition. Whether this factor suf®ces for a complete explanation of our results and the previous results is an issue we take up in the General discussion. As in the foregoing experiments the present experiment also allows us to assess the in¯uence of a general warning e€ect on RT. We obtained results similar in pattern to those found in the previous experiments. In the no-prime condition there was a main e€ect of delay: the mean RT in the 250 ms delay condition was 14 ms smaller than the mean RT in the 150 ms delay condition.

General discussion Visual information processing and visual perception are not instantaneous all-or-none processes but processes that develop gradually and progressively over time. It is, therefore, quite conceivable that the initial stages of

102 Table 6 The general warning e€ects (de®ned as the di€erence in RT between the 150 ms and 250 ms delay conditions) and the e€ect of a prime in the ®ve experiments

General warning e€ect

Experiment Experiment Experiment Experiment Experiment

1 2 3 4 5

E€ect of prime (N-T)-(T-T)

Prime (N-T)

No Prime (T-T)

Mean

Delay 150

Delay 250

Mean

)9 )8 0 2 )5

17 12 15 10 14

4 2 7 6 4

16 17 3 )9 )46

42 37 18 )1 )27

29 27 10 )5 )36

processing do not produce as precise discriminatory detail as later stages of processing. If so, it is also conceivable that a stronger condition holds, that the initial period of presentation of a stimulus can be replaced by other visual information without a€ecting the course of processing of that stimulus. That is, there might be a period of time at the onset of stimulation in which only the presence of information rather than its content is required for processing to proceed. Ho€man (1975), Eriksen et al. (1990) and Shi€rin et al. (1996) reported studies that purportedly provided evidence for this ``perceptual inertia'' hypothesis. The present series of ®ve experiments paint a di€erent picture. Although not ruling out the existence of (some) perceptual inertia, they provide evidence that the results are explicable in terms of several standard processing mechanisms: forward masking, selective attention to location, and ``warning period'' duration. Table 6 summarizes the main results of our experiments. The left panel of the table shows for each of the ®ve experiments the general warning e€ects, de®ned as the di€erence in mean RT between the 250 ms delay and 150 ms delay conditions (a positive value indicates that RTs decrease when the foreperiod increases; a negative value indicates the opposite). Inspection of this part of the table shows that there is a positive, i.e., facilitating, general warning e€ect in the T-T conditions but not in the N-T conditions: there is about 14 ms (signi®cant) advantage for the longer warning period, when there is no prime. There is a (non-signi®cant) 4 ms advantage for the shorter warning period when there is a prime preceding the target for 100 ms. Because of a speed-accuracy trade-o€, this advantage is somewhat underestimated (see Tables 1±5). The positive delay e€ects in the T-T conditions and the negative delay e€ects in the N-T conditions can be explained in the following way. In all our experiments there was a ®xation cross that disappeared 150 ms or 250 ms before T-T or N-T presentation. At the start of the trials that ®xation cross was attended. For performing the required task the participant's attention has to disengage from the position of the ®xation cross. It is well known that the disengagement of attention is a time-consuming process (see, e.g., Posner & Raichle, 1994). It is, therefore, reasonable to assume that, after disappearance of the ®xation cross, i.e., in the time interval when attention is disengaging, the susceptibility for new visual information gradually increases. For the T-T conditions this entails that the target, T, is more

eciently processed the later it appears. This will result in shorter latencies with longer delays, as observed. For the N-T conditions this entails that the prime, N, is more eciently processed the later it appears. Because of interference, this can result in longer latencies with longer delays, as generally observed. In the right panel of the table for each of the ®ve experiments are given the estimated amounts of ``dead time'' in the two delay conditions, 150 ms and 250 ms (a negative value indicates the presence of a period of dead time and a positive value indicates the opposite). Experiments 2, 3, 4, and 5, taken together, form the orthogonal combination of the factors (a) one prime versus two primes (the presence or absence of an e€ect of selective attention, respectively) and (b) letter primes versus dot primes (a large amount of forward masking versus a low amount of forward masking, respectively). An ANOVA in which the results of these four experiments were combined showed that both factors a€ected the size of the priming e€ect: the use of one instead of two primes (selective attention) resulted in a 39 ms decrease of the priming e€ect (from an average priming e€ect of 18.5 ms in Experiments 2 and 3 to an average e€ect of ±20.5 ms in Experiments 4 and 5; see Table 6), F(1, 28) ˆ 66.0, p < 0.001; the use of a dot prime instead of a letter prime (reduction of forward masking) resulted in a 24 ms decrease of the priming e€ect (from an average priming e€ect of 11 ms in Experiments 2 and 4 to an average e€ect of ±13 ms in Experiments 3 and 5; see Table 6), F(1, 28) ˆ 24.8, p < 0.001. The e€ects of selective attention and masking on the priming e€ect proved to be additive: the second order interaction between the factors number of primes (one versus two), prime type (dot versus letter), and presence versus absence of a prime failed to reach signi®cance (p ˆ 0.13). Thus, the results of these four experiments are consistent with, and can be explained in terms of three standard phenomena: warning period (14 ms, the mean of the values in the second column of Table 6), selective attention (±39 ms), and forward masking (24 ms). The results of the experiment of Ho€man (1975), which led to the introduction of the explanatory concept of perceptual inertia, are consistent with those from the present Experiment 5. In both cases, one may assume an advantage conferred by selective attention to location (see van der Heijden, 1992, Chapter 4, for a summary). When the prime provides a cue to location, it allows attention to be moved sooner to that location. In the Ho€man studies, when the display contained just one

103

item (in a circle of locations) it may be presumed that attention moves quickly to the relevant (only) location, and little advantage of the prime is seen (as found); on the other hand, with eight ®lled locations, even with a bar marker indicating the relevant location, it may take some time for attention to be directed to the relevant location, time during which the identity of the details of the character at that location are not important. In conclusion, Ho€man's (1975) ®ndings can rather parsimoniously be explained in terms of an e€ect of selective attention (the prime acted as a location cue), in combination with the absence of forward masking (due to the use of a subset of the target features as a prime stimulus) and an e€ect of warning foreperiod duration. With respect to the latter factor, note that in Ho€man's study the warning foreperiod duration was smaller in the T-T condition than in the N-T condition (see also Fig. 1). Our present results indicate that if Ho€man had used identical warning foreperiods in these two conditions, the size of the alleged perceptual inertia e€ect would have been smaller. Why did Shi€rin et al. (1996) obtain evidence for perceptual inertia? Two explanations are worth consideration. First, in Shi€rin et al.'s Experiment 1a a cue was presented 185 ms in advance of any letters, then the prime display was presented for 150 ms and ®nally the target display was presented until response. With such a sequence of exposure times subjects may make directed saccadic eye movements towards the cue position (see, e.g., Rayner, Slowiaczek, Clifton, & Bertera, 1983, for information on the latency of eye movements). Just before, during and after a saccadic eye movement visual sensitivity is reduced, a phenomenon called saccadic suppression (see, e.g., Latour, 1966; Volkman, 1962; Zuber, Stark, & Lorber, 1996). Thus, the sequence of exposure times used by Shi€rin et al. may have allowed some real ``perceptual inertia'', due to saccadic suppression. It should be noted, however, that an explanation in terms of saccadic suppression is much less likely for other experiments reported by Shi€rin et al., in which the presentation of the cue was delayed over a wide range without much altering the pattern of results. The most likely explanation for the results obtained by Shi€rin et al. (1996) was mentioned as an alternative explanation by those authors in a footnote (Shi€rin et al., p. 239, footnote 1). In their experiments the prime stimulus could sometimes be the alternative target, and then switch after 150 ms to the imperative target, to which the response was to be made (Condition T2-T). This feature of the experiment might have led the participants to delay their responding until they were surer concerning the identity of the imperative target. Such a strategy will show up as a somewhat delayed responding in their T-T and T2-T conditions, relative to their N-T condition, (whatever the direction of di€erence between these). Thus, this ``cautious'' strategy is capable of explaining Shi€rin et al.'s evidence for perceptual inertia, i.e., the ®nding that in the T-T condition the mean RT was much less than 150 ms smaller than in the N-T

condition in which the target appeared 150 ms later, even in the presence of an e€ect of forward masking. The studies of Eriksen et al. (1990) showed no difference between RT in their N-T, T-T, and T2-T conditions. In each of these cases the prime occurred in an initially unattended location, followed by the target (and a second location cue) after 50 ms. The hypothesis of perceptual inertia is founded upon the assumption that automatic processing of the unattended prime ought to have occurred, with the results loaded into a response producing mechanism that (in the absence of perceptual inertia) would cause compatibility e€ects. It may be, however, that the ®rst location cue removes attention from the second location, and processing that occurs there (if any) does not enter a response producing mechanism and, therefore, does not produce compatibility e€ects.

Concluding remarks Perceptual inertia could be described as the hypothesis that there is a brief period of time after onset of visual material at a given location during which general processing occurs but during which details of the information at that location are not extracted; as a corollary, these details could be altered during this period without a€ecting the general time course and quality of processing. 1. Perceptual inertia is a conceptually reasonable idea, which could be justi®ed with several models of information processing. 2. Previous studies providing evidence interpreted as favoring perceptual inertia were aimed at other topics and issues, and were not well designed to study perceptual inertia per se. 3. The ®ve studies in this article were aimed at testing the existence of this construct in a direct fashion. Evidence for perceptual inertia would have been faster responding measured from the onset of the target when the target is preceded by a noise character. a. The results from Experiments 1, 2, and 3 which controlled selective attention and other variables, did not show this pattern. The results were consistent with the hypothesis that forward masking is produced by the leading noise character, slowing responding. b. Experiments 4 and 5 introduced the possibility that attention to location would selectively bene®t the condition with the leading noise character. The results were consistent with the hypothesis that this indeed occurred. 4. The present studies do not rule out the possibility that perceptual inertia exists, because the most relevant test compares a condition with forward masking with another without forward masking. All that can be said with certainty is that any perceptual inertia is outweighed by the masking factor.

104

5. The results from our experiments indicate that evidence for perceptual inertia can be explained parsimoniously through the operation of three wellstudied standard factors, warning period duration, forward masking, and selective attention to location. Acknowledgements The authors thank Joke M. P. Baas for running part of the experiments and Peter Dixon, Charles Folk and Roger Remington for their comments on a version of this article.

References Coltheart, M. (1980). Iconic memory and visible persistence. Perception and Psychophysics, 27, 183±228. Eriksen, B. A., & Eriksen, C. W. (1974). E€ects of noise letters upon the identi®cation of a target letter in a nonsearch task. Perception and Psychophysics, 16, 143±149. Eriksen, C. W., Webb, J. M., & Fournier, L.R. (1990). How much processing do nonattended stimuli receive? Apparently very little, but.... Perception and Psychophysics, 47, 477±488. Ho€man, J. E. (1975). Hierarchical stages in the processing of visual information. Perception and Psychophysics, 18, 348± 354. Kahneman, D. (1968). Method, ®ndings and theory in studies of visual masking. Psychological Bulletin, 70, 404±425. Latour, P. (1966). Visual threshold during eye movements. Vision Research, 2, 261±262.

Neisser, U. (1967). Cognitive psychology. New York: AppletonCentury-Crofts. Posner, M. I., & Boies, S. J. (1971). Components of attention. Psychological Review, 78, 391±408. Posner, M. I., & Raichle, M. E. (1994). Images of the mind. New York: Scienti®c American Library. Posner, M. I., Snyder, C. R. R., & Davidson, B. J. (1980). Attention and the detection of signals. Journal of Experimental Psychology: General, 109, 160±174. Rayner, K., Slowiaczek, M. L., Clifton, C., & Bertera, J. H. (1983). Latency of sequential eye movements: implications for reading. Journal of Experimental Psychology: Human Perception and Performance, 9, 912±922. Rohrbaugh, J. W., & Eriksen, C. W. (1975). Reaction time measurement of temporal integration and organization of form. Perception and Psychophysics, 17, 53±58. Shi€rin, R. M., Diller, D., & Cohen, A. (1996). Processing visual information in an unattended location. In A. F. Kramer, M. G. H. Coles & G. D. Logan (Eds.). Converging operations in the study of visual selective attention. Washington, DC: American Psychological Association. Sperling, G. (1960). The information available in brief visual presentations. Psychological Monograph, 74(11) (whole no. 498). Van der Heijden, A. H. C. (1992). Selective attention in vision. London: Routledge. Volkman, F. (1962). Vision during voluntary saccadic eye movements. Journal of the Optical Society of America, 52, 571±578. Zuber, B., Stark, L., & Lorber, M. (1966). Saccadic suppression: Elevation of visual threshold associated with saccadic eye movements. Experimental Neurology, 16, 65±79.