Iacoboni M., Rayman J., Zaidel E. (1997) Does the previous trial affect lateralized lexical decision? Neuropsychologia, 35: 81-88

Does the Previous Trial Affect Lateralized Lexical Decision?

Marco Iacoboni1,2, Jan Rayman2, Eran Zaidel2

1Division of Brain Mapping, Department of Neurology, UCLA School of Medicine

2Department of Psychology

University of California, Los Angeles

Address correspondence to :

Marco Iacoboni, M.D. Ph.D.
Reed Neurological Research Center
Department of Neurology, UCLA School of Medicine
710 Westwood Plaza
Los Angeles, California 90095
e-mail: iacoboni@loni.ucla.edu
Tel (310) 206-3992
Fax (310) 794-7406

We thank Elicia David and Krista Schendel for research assistance. This work was supported by NIH NS 20187 and by NIMH RSA MH 00179.

Abstract

We investigated the effect of previous trial variables on performance in the current trial in a lexical decision task with unilateral presentation of one letter string or bilateral simultaneous presentation of two different letter strings, one cued to be processed (target) and the other uncued, to be ignored (distractor). The variables included correctness of the previous trial, visual hemifield and wordness of the previous trial, and presentation mode of the previous trial (unilateral or bilateral). An incorrect response on the previous trial enhanced the accuracy in the current trial only in the left visual field (LVF). A previous LVF target produced faster correct responses to LVF targets in the current trial and LVF word recognition was more accurate when the previous LVF target was a word rather than a nonword. Target processing in the current trial was not inhibited or facilitated if the target belonged to the same response category as the unattended stimulus in the previous trial (absence of "negative priming").

Taken together, our data suggest that previous trial effects in lateralized lexical decision are stronger for word decisions in the LVF, and may account for the inconsistency of the wordness effect in the LVF across different lateralized lexical decision experiments. Our data also suggest that behavioral laterality experiments are well advised to use random sequences that change across subjects in order to minimize previous trial effects.

Introduction

Hemifield tachistoscopic presentations of verbal and non verbal stimuli are frequently used to investigate hemispheric competence in the normal brain. A visual hemifield (VF) advantage in accuracy and/or latency in a given task is taken to reflect a superiority of the contralateral hemisphere for that task. However, VF asymmetries for the same task or population are notoriously variable in magnitude and occasionally they can even vary in direction. The variability in VF asymmetries in hemifield tachistoscopic experiments limits the interpretation of the empirical data in terms of hemispheric competence in the normal brain. A simple and increasingly used methodology to reduce the variability in VF asymmetries is the simultaneous bilateral presentation of two different stimuli, one as the target to be processed, the other as a distractor, to be ignored by the subject. Using this method, referred to henceforth as the "bilateral presentation mode", it is possible to obtain larger, more significant, and more reliable VF asymmetries in lateralized experiments with verbal and non verbal stimuli, as repeatedly shown [3-8, 16]. The bilateral presentation mode may produce greater VF asymmetries by maximizing hemispheric independence not only of strategy but also of resources, operationally indexed by a significant response hand by VF interaction [19], by means of simultaneously engaging both hemispheres in the automatic processing of the target in one VF and the distractor in the contralateral VF [8, 16].

However, even the bilateral presentation mode may introduce confounding factors, such that VF asymmetries may not reflect the real competence of the two hemispheres in the domain of interest. Instead, VF asymmetries may also reflect facilitatory and/or inhibitory context effects due to processing the preceding trials. For example, the bilateral presentation mode may produce the so-called "negative priming effect", well-known from free-vision perceptual tasks. If the unattended stimulus in the previous trial belongs to the same response category as the target in the current trial, then an inhibition can occur [1, 2]. However, this inhibition is seen only when the target in the current trial is simultaneously presented with a distractor. When the target in the current trial is presented alone, then a facilitation can occur instead [1, 2]. A lateralized version of the negative priming effect has been described recently [10]. The standard interpretation of this phenomenon in perception is in terms of response inhibition and/or facilitation of stimuli belonging to the same category [1, 2, 10]. If such priming effects are effective with bilateral presentation in lexical decision, then we should observe an inhibition of target processing for sequential pairs of bilateral trials when the distractor in the previous trial has the same lexical status as the target in the current trial, as well as a facilitation of target processing for bilateral (previous) trials followed by unilateral (current) trials when the distractor in the previous trial has the same lexical status as the target in the current trial. These priming effects, if present, could produce serious confounding factors affecting the overall VF asymmetries observed in lexical decision experiments, such that the interpretation of these asymmetries in terms of hemispheric competence for verbal stimuli would be problematic.

The goal of the present paper is to investigate whether such priming effects influenced the performance in the current trial in a lateralized lexical decision experiment with unilateral and bilateral trials designed to investigate the differential competence and independence for word recognition in the two hemispheres [8]. If this re-analysis shows that previous trial variables influence the performance in the current trial in lateralized lexical decision experiments, then behavioral laterality experiments should include a complete counterbalancing of previous and current trial variables. To the best of our knowledge, this is the first time that an analysis of previous trial variables on the performance in the current trial is reported in lateralized lexical decision.

Other influences of the previous trial on the performance in the current trial, more general and not related to the bilateral presentation mode per se, have been described in the literature. A general rule in speeded choice reaction time tasks is that if the same stimulus occurs in two consecutive trials, responses are faster than if the stimuli occurring in two consecutive trials are different (repetition effect). This phenomenon seems to be subserved by a transient strengthening of the stimulus-response mapping that shortcuts the response-selection stage [14]. This repetition effect calls for the use of extensive lists of stimuli in lateralized lexical decision experiments, to avoid spurious effects produced by repeated presentations of the same stimulus in two consecutive trials, favoring by chance one VF and not the other (unless the repeated presentations are completely counterbalanced between the two VFs), thus altering the VF difference and affecting its validity as an index of differential verbal hemispheric competence.

Another lateralized "previous trial" phenomenon has been described recently [11]. It was shown that both the lateralization and the response accuracy of the previous trial can affect performance in the current trial in a lateralized syllable identification task and these effects have been interpreted in terms of hemispheric arousal systems. Two methodologically important questions related to these findings are (1) the generalizability of these effects to other tasks and (2) their magnitude. If previous trial effects extend to other tasks, such as lexical decision, they may alter the current trial performance in such a way as to again make difficult the interpretation of VF asymmetries in terms of the lexical competencies of the two hemispheres. However, this would be empirically important only if the magnitude of these previous trial effects produces reliable changes in the performance in the current trial. If the overall magnitude of the previous trial effects accounts only for a small portion of the experimental variance, then we may conclude that previous trial effects in lateralized lexical decision tasks are negligible and require no counterbalancing.

In a recent lexical decision experiment we used unilateral and bilateral presentation of different stimuli. Each stimulus was presented only once in each experimental session, to circumvent repetition effects. In the bilateral presentation mode one stimulus was attended and the other unattended [8]. The bilateral presentation mode was used to increase VF asymmetries [3-6] and to maximize hemispheric independence, indexed by a significant response hand by VF interaction [19]. In principle, the use of unilateral and bilateral presentation modes in the same experiment may have produced spurious negative priming effects, i.e., the wordness of the unattended stimulus in the previous trial may inhibit or facilitate the processing of the target in the current trial. Moreover, if the previous trial effects described in the lateralized syllable identification task [11] were also effective in lexical decision, then the correctness of response in the previous trial and the VF of the previous target may have altered the VF asymmetry of the performance in the current trial. To address these questions we analyzed with a post hoc procedure the effect of the previous trial on the performance in the current trial in our lexical decision experiment with unilateral and bilateral presentation modes. We did not design an experiment specifically on the previous trial in lateralized lexical decision because we wanted to test whether possible previous trial effects may "contaminate" the performance in the current trial in a standard lateralized lexical decision task that maximizes hemispheric independence and perceptual asymmetries. This approach may help us understand whether lateralized lexical decision experiments require not only the counterbalancing of current trial variables but also the counterbalancing of previous trial variables.

Methods

Subjects. Twenty-four undergraduate UCLA students participated in this experiment. All the subjects were strongly right-handed as determined by a handedness inventory, had no left-handed relatives, and had not spoken or understood any language except English until at least the age of six. All subjects reported normal or corrected-to-normal vision in both eyes and no history or evidence of neurological insult. The subjects received course credit for their participation.

Apparatus. Subjects were seated in a dimly lit room at a distance of 57 cm from a high resolution RGB color monitor of a MacIntosh IIsi computer, with their chins in a chinrest, their eyes aligned with the fixation cross in the middle of the screen, and index and middle fingers poised on keys of the computer keyboard placed symmetrically at midline and aligned vertically (g and b for the left hand, j and n for the right hand). A green label with w or n on it, indicating respectively buttons for words and for nonwords, was placed on each key. A computer software package for MacIntosh, MacProbe, was used to present stimuli and to record responses.

Procedure. A fixation cross was displayed during the entire experiment. A warning tone sounded 750 msec before the presentation of stimuli. Displays, horizontal lowercase letter strings, were presented for 120 msec and they were black on a gray background. The innermost edge of the letter string appeared 1.5 degrees to the right and/or to the left of the fixation cross. The strings subtended from 1.5 to 3.0 degrees of visual angle. In half of the trials only one letter string in either the left (LVF) or the right visual field (RVF) was presented (unilateral trials); in the other half two different letter strings of the same length appeared in each visual field, one as a target and the other one as a distractor (bilateral trials). An arrow indicating the target was displayed simultaneously with the letter strings, with the inner edge at 0.6 and the outer edge at 0.9 degrees from the fixation cross, both in unilateral and bilateral trials. The subject's task was to decide whether the letter string indicated by the arrow was a word or not by pressing the key with the corresponding label.

The experiment was repeated thrice for each subject. Subjects were instructed to use left hand (Lh) only, right hand (Rh) only, and both hands simultaneously (Bh). The order of hand responses was counterbalanced between subjects. Each subject participated in a practice session before each response hand condition. For each response hand condition each subject received 192 trials divided into three blocks of 64. Half of the subjects were instructed to use the index finger for words and the middle finger for nonwords, and the other half were instructed to use the index finger for nonwords and the middle finger for words.

Stimulus materials. Stimuli were 288 letter strings three, four, five, and six letter long; 144 were words and 144 pronounceable orthographically regular nonwords that were matched for length. The full list of words and nonwords used for the experiment is presented elsewhere [8].

All the items and their linguistic effects were balanced between subjects by the creation of twelve different lists. This was implemented by rotating the 288 letter strings in all the twelve possible combination of left and right VFs, unilateral and bilateral presentation, wordness of the target, and, only for bilateral trials, wordness of the distractor. The twelve lists were balanced between subjects. The order of presentation of the trials composing each list, however, was randomly generated by the computer program. So, even subjects receiving the same lists were not exposed to the same order of presentation.

Data analysis. Analyses of variance (ANOVAs) with repeated measures were performed for percentage of correct trials and medians of reaction times (RTs) of correct responses as dependent variables. Post hoc comparisons incorporated the Bonferroni-Dunn correction. We also estimated the RTs variance accounted for by each source in our model as well as the variance in VF asymmetries accounted for by the previous VF and by individual differences, using a stepwise multiple regression approach [15, 17].

Stimulus lists, sex, and response finger condition were between-subject factors, and VFs of the current trial (CurrVF), VFs of the previous trial (PreVF), and response accuracy for the previous trial were within-subject factors in preliminary ANOVAs. Since in no analysis were the main effects or interactions involving response finger condition, sex, and stimulus lists ever significant, these constraints were removed from subsequent ANOVAs.

Results

A detailed presentation of the results of the performance in the current trial (trial n) is provided elsewhere [8]. The main results were an RVF advantage, a wordness effect (words processed faster and more accurately than nonwords), and a presentation mode effect (bilateral displays processed more slowly and less accurately than unilateral displays). In addition, several interactions were observed. In particular, bilateral displays enhanced VF asymmetries, affected word processing but not nonword processing, and did so mostly in the LVF compared to the RVF. Bilateral displays also showed a "lexicality priming" effect, that is, word targets were processed faster with word distractors, whereas processing of nonword targets was not affected by the wordness of the distractor. In the following sections we will present only effects and interactions involving the relationships between the previous and the current trial variables, results that are not reported anywhere else. The experimental variables and their possible interactions were not analyzed all at once because the number of data points contributing to the mean in each cell of the repeated measures design would have been too small. As a general rule, we did not perform any analysis in which the number of data points in each cell was below 10.

Effect of correctness and VF of targets of the previous trial. The effect of correctness and VF of targets of the previous trial on the performance in the current trial was tested in 2 (Previous Trial: correct, incorrect) by 2 (Previous VF: left, right) by 2 (Current VF: left, right) ANOVAs with accuracy and RTs as dependent variables.

Accuracy. A Current VF by Correctness of the Previous Trial interaction was observed in accuracy, F(1, 23)=9.340, p<.01. Planned comparison showed that LVF targets were processed more accurately, F(1, 23)=7.158, p<.01, if preceded by an incorrect previous trial (75.3%), compared to a correct previous trial (73%). RVF targets were processed slightly more accurately when preceded by a correct trial (86.2%) compared to an incorrect trial (84.9%) but the difference was not statistically significant (Figure 1). No other main effects or interactions were significant.

Figure 1 Current decisions in the LVF were significantly more accurate when the previous trial was correct. ** for p<.01

RTs. An interaction Current VF by Previous VF was observed in RTs, F(1, 23)=13.990, p<.01. LVF targets were processed faster, F(1, 23)=11.729, p<.01, when preceded by a LVF target (806 msec) compared to a RVF target (842 msec). RVF targets were processed slightly faster when preceded by a RVF target (750 msec) compared to a LVF target (769 msec), but the difference fell short of significance, F(1, 23)=3.478, p<.08 (Figure 2). No other main effects or interactions were significant.

Figure 2 RTs for current LVF targets were significantly shorter when the previous target was in the LVF. ** for p<.01

Effect of the wordness of the previous target. The effect of the wordness of the previous target was tested in a 2 (Previous Trial: correct, incorrect) by 2 (Previous VF: left, right) by 2 (Current VF: left, right) by 2 (Previous Target Wordness: word, nonword) by 2 (Current Target Wordness: word, nonword) ANOVA.

Accuracy. The Current VF by Correctness of the Previous trial interaction, previously described, was observed again. The Previous VF by Current VF by Current Target Wordness interaction approached significance, F(1, 23)=3.161, p<.09. Since the same interaction reached a full significance in RTs we will discuss it in the RTs section. A four-way interaction involving Previous VF, Current VF, Previous Target Wordness, and Current Target Wordness was observed for accuracy, F(1, 23)=16.934, p<.001. Planned comparisons showed that word targets in the LVF were processed more accurately, F(1, 23)=15.241, p<.001, when preceded by a LVF word target (77.3%) compared to a LVF nonword target (70.8%) (Figure 3). No other comparisons showed significant differences. No other main effects or interactions were significant.

Figure 3 Current decisions for words in the LVF were significantly more accurate when the previous LVF target was a word. *** for p<.001

RTs. A main effect of the Previous Target Wordness, F(1, 23)=16.934, p<.001 was observed in RTs. Word targets in the previous trial produced a faster RT in the current trial (783 msec) compared to nonword targets (803 msec). The interaction Current VF by Previous VF, previously described, was observed again. A Current VF by Previous VF by Current Target Wordness interaction was observed in RTs, F(1, 23)=11.392, p<.01. Word targets in the LVF were processed faster, F(1, 23)=55.534, p<.001, if preceded by a LVF target (762 msec) compared to a RVF target (816 msec). RVF word targets were also processed faster, F(1, 23)=14.005, p<.01, when preceded by a RVF target (680 msec) compared to a LVF target (707 msec) (Figure 4). No other comparisons showed significant differences. A Previous VF by Current VF by Previous Target Wordness by Current Target Wordness interaction was significant in RTs, F(1, 23)=5.132, p<.05. The nature of this interaction is similar to the one described for accuracy, and we will not discuss it further. No other main effects or interactions were significant.

Figure 4 RTs for current word targets were significantly shorter when the previous VF target was the same as the current VF target. ** for p<.01, *** for p<.001

Effect of presentation mode. The effect of presentation mode was tested in a 2 (Previous Trial: correct, incorrect) by 2 (Previous VF: left, right) by 2 (Current VF: left, right) by 2 (Previous Trial Presentation Mode: unilateral, bilateral) by 2 (Current Trial Presentation Mode: unilateral, bilateral) ANOVA. No significant main effects or interactions involving presentation mode and previous trial were observed for either accuracy or RT.

"Negative priming" and "lexicality priming". We tested for the occurrence of a "negative priming" effect and its possible interaction with the "lexicality priming" effect in a 2 (Previous Distractor Wordness: word, nonword) by 2 (Previous Target Wordness: word, nonword) by 2 (Current Distractor Wordness: word, nonword) by 2 (Current Target Wordness: word, nonword) ANOVA for sequential pairs of bilateral trials only. No significant interactions were observed for either accuracy or RT. In addition, we tested the "facilitating version" of the "negative priming" effect in a 2 (Previous Distractor Wordness: word, nonword) by 2 (Previous Target Wordness: word, nonword) by 2 (Current Target Wordness: word, nonword) ANOVA for sequential pairs of bilateral (previous) trial - unilateral (current) trial only. Again, we did not observe any significant interaction. In particular, we did not observe an inhibition of current target processing when the distractor of the previous target had the same lexical status in bilateral pairs, and we did not observe a facilitation of current unilateral target when the previous distractor had the same lexical status as the current target.

Sources of variance in the repeated measures design. To estimate the percentage of variance of various factors in our repeated measures design we used a stepwise multiple regression approach [15, 17]. Independent factors entered were Subjects, Current VF, Correctness of the Previous Trial, Previous VF, Current VF by Previous VF, Current VF by Correctness of the Previous Trial, Correctness of the Previous Trial by Previous VF, and Current VF by Correctness of the Previous Trial by Previous VF. The dependent variable was RT. The largest percentage of variance came from individual differences (36.7%; p<.001). This is a well known phenomenon in laterality research and, more generally, in behavioral investigations. The Current VF source accounted for 7.9% of the variance (p<.001), whereas the Current VF by Previous VF interaction source accounted only for an additional 1.5% of the variance (not significant). All the remaining sources accounted for even smaller percentages of the variance.

VF asymmetries and the previous VF. To estimate the percentage of variance in VF asymmetries due to the previous VF and due to individual differences, which were the two factors accounting for the largest portion of variance according to our previous stepwise regression analysis, we again used a stepwise multiple regression approach [15, 17]. The two factors entered were Previous VF and Subjects, and the dependent variable was the VF asymmetry in RT (LVFrt - RVFrt) for trials preceded by LVF targets and for trials preceded by RVF targets. The previous VF target accounted for 18.2% (p<.01) of the total variance in VF asymmetries, whereas individual differences accounted for 17.7% (p<.01) of the total variance in VF asymmetries. Remarkably, the previous VF target accounted for slightly larger amount of variance in VF asymmetries than did individual differences.

Discussion

The rationale of the present study was to investigate the role of a variety of possible previous trial effects on the performance in the current trial. Our concern was that the bilateral presentation mode, which effectively produces larger VF asymmetries in hemifield tachistoscopic experiments, may also produce "spurious" priming effects, such as the negative priming effects described in free vision perceptual tasks [1, 2] and more recently in a lateralized task [10]. These priming effects, tentatively explained as response suppression phenomena [1, 2, 10], can be regarded as "spurious" if they produce greater VF asymmetries for reasons not related to differential processing properties of the current target of either of the two hemispheres. To the best of our knowledge, negative priming effects have never been investigated in a lexical decision task. Our data seem to suggest that these priming effects are not strongly effective in lateralized lexical decision tasks. We did not observe either the inhibition of target processing in current bilateral trials (when the previous distractor had the same lexical status as the current target) or the facilitation of target processing in current unilateral trials (when again the previous distractor had the same lexical status as the current target). However, the nature of the cue adopted in our experiment to indicate the target (an arrow between fixation and target) seems to be different from the cues used in previous perceptual experiments on negative priming [1, 2]. In particular, in our experiment the target location is exogenously cued in bilateral trials but both exogenously and endogenously cued in unilateral trials (the word itself may be regarded an endogenous cue). Moreover, in our experiment we interspersed unilateral and bilateral trials. We have commented elsewhere that this may have especially altered the processing of unilateral trials [8]. Thus, the findings on previous trial effects in this study may not generalize to experiments in which unilateral and bilateral trials are presented separately. Therefore, further data are needed to completely exclude the occurrence of negative priming effects in lexical decision experiments with bilateral displays.

More generally, we did not observe any interaction of current and previous trial variables with presentation mode. This rules out the possibility of "spurious" previous trial effects in the consistently observed enlargement of VF asymmetries produced by the bilateral presentation mode and reaffirms the validity of the conclusion that bilateral presentation mode enhances hemispheric asymmetries in the current trial [3-6] probably by facilitating hemispheric independence [7, 8, 16]. By the same token, the absence of previous trial effects in the "lexicality priming" effect suggests that this latter effect is a reliable "current trial" phenomenon [8] and deserves to be further investigated.

We observed four types of previous trial effects in our lateralized lexical decision experiment. First, there was differential effectiveness of previous trial response accuracy in the two VFs: An incorrect response in the previous trial significantly enhanced the performance in the current trial in the LVF, whereas the correctness of the previous trial did not significantly affect the performance in the current trial in the RVF. This result is consistent with the evidence that the right hemisphere has better abilities in error monitoring than the left hemisphere [18]. These differential error monitoring abilities seem to be effective at the late stage of response programming [18], which is in line with the other previous trial effects observed in this study.

The second previous trial effect observed in this investigation consists of a faster processing for current LVF targets when the previous target was in the same LVF. Such a facilitation, even though not significant, was also observed for current RVF targets when the previous target was in the RVF. These results may suggest the existence of two parallel hemispheric activation systems during lexical decision, each affecting the performance of "its" own hemisphere. The tendency for a more robust effect of the previous trial in the LVF may be simply due to the task difficulty and greater VF asymmetry produced by the bilateral display. The LVF may have had more resource limitations and therefore more room to be affected by changes in activation as well as more room to improve its performance. The RVF already had a high accuracy and may have had resources to spare, thereby buffering it from the effects of arousal produced by previous trials. At variance with our results, others [11] have observed that performance in the current trial in a lateralized syllable identification task was better in both VFs when the previous stimulus was presented in the LVF rather than in the RVF, and that performance in the current trial was also better in both VFs when the response to the LVF previous trial was correct. These findings were interpreted as suggesting an asymmetry of hemispheric arousal systems, the right hemisphere arousal system being involved in generating and sustaining bilateral arousal independently from the adequacy of processing of the previous trial.

The inconsistency between our results and the previous trial effects in lateralized syllable identification [11] can be reconciled assuming that previous trial effects are task-specific (see ref. 12 for other previous trial effects in perceptual tasks). If this is true, then it is unlikely that previous trial effects reflect one or two parallel nonspecific hemispheric arousal systems. The previous trial may produce its effect at a later stage, for instance at the stage of task-specific response selection processes. This interpretation is supported by the third and fourth previous trial effects observed in the present investigation. Word targets were processed significantly faster in each VF when the previous target was a word in the same VF. Further, decisions for word targets in the current LVF were significantly more accurate when the previous LVF target was a word. It seems to us unparsimonious to argue that the activation of a nonspecific arousal system may selectively produce effects involving a specific lexical category (word targets). On the contrary, there is evidence for independent and parallel word and nonword recognition processes and response programming in the two hemispheres during lexical decision [8, 13]. Our data suggest that these processes can be independently activated during hemifield tachistoscopic lexical decisions and this may produce differential amounts of previous trial effects in the two VFs and for word and nonword recognition. The independent activation of word and nonword recognition processes would make signal detection analysis, which assumes a comparison between a signal population (words or nonwords) and a noise population (nonwords or words, respectively) along a single monotonic scale, an inappropriate approach for lateralized lexical decision tasks [8, 13]. Our results are also consistent with the hypothesis that the independent modules in the two hemispheres have different dynamics. In particular, right hemisphere modules may change more widely but more slowly that left hemisphere modules, and may be more sensitive to context effects, such as the previous trial effect.

Finally, although the correctness of the previous trial, the previous VF, and the wordness of previous targets can all affect performance in the current trial, as evidenced by the previous trial effects observed in this experiment, the largest interaction of these previous trial variables with the VF in the current trial accounted only for 1.5% of the RT variance. This suggests that previous trial effects are generally small, at least in lateralized lexical decisions. However, when VF asymmetries in RTs are considered, the previous VF target accounts for a large amount of variance, even slightly larger than the individual differences. This outcome suggests that the independent word and nonword recognition systems in the left and right hemispheres, when activated by the presentation of a given stimulus in the contralateral VF in trial n - 1, are subsequently facilitated in trial n. According to our data, this seems to be especially effective for word recognition in the right hemisphere (as indexed by word decisions in the LVF). This hypothesis may explain why the "wordness effect" in the LVF (word decisions more accurate and faster than nonword decisions) is so inconsistently observed in lateralized lexical decisions. If word processing in the right hemisphere, having more "momentum", is highly sensitive to previous trial variables, which may not be well controlled and counterbalanced in lateralized lexical decision (counterbalancing being usually explicitly controlled only for the current trial), then we can expect a large variability of word decisions, hence a large variability of the wordness effect in the LVF across tasks. This may be systematically investigated, re-analyzing the previous trial effects in those published lateralized lexical decision experiments that show inconsistency in terms of wordness effect in the LVF. Note that the lack of a wordness effect in the LVF is a strong challenge to the traditional models of word recognition (for a discussion of this point, see ref. 8). If the selective role of previous trial variables in influencing word recognition processes in the LVF will be confirmed, then only a lateralized lexical decision experiment in which both current trial and previous trial variables are counterbalanced may reliably assess the wordness effect in the LVF.

In conclusion, previous trial effects in hemifield tachistoscopic experiments reflect the differential dynamics of independent and specific modules in the two hemispheres. These modules are task- and stimulus-specific (e.g. word recognition modules in lexical decision). Some of them show greater momentum across trials and thus more pervasive previous trial effects.

References

1. Allport, A. Visual attention. In Foundations of Cognitive Science, M.I. Posner (Editor), pp. 631-682. MIT Press, Cambridge, MA, 1989.

2. Allport, D.A., Tipper, S.P. and Chmiel, N.R.J. Perceptual integration and postcategorical filtering. In Attention and Performance XI, M.I. Posner and O.S.M. Marin (Editors) pp. 107-132. Lawrence Erlbaum, Hillsdale, N.J., 1985.

3. Boles, D.B. Hemispheric interaction in visual field asymmetry. Cortex 19, 99-113, 1983.

4. Boles, D.B. Reaction time asymmetry through bilateral vs. unilateral stimulus presentation. Brain Cogn. 6, 321-333, 1987.

5. Boles, D.B. What bilateral displays do. Brain Cogn. 12, 205-208, 1990.

6. Boles, D.B. An experimental comparison of stimulus type, display type, and input variable contributions to visual field asymmetry. Brain Cogn. 24, 184-197, 1994.

7. Hines, D. Independent functioning of the two cerebral hemispheres for recognizing bilaterally presented tachistoscopic visual-half-field stimuli. Cortex 11, 132-143, 1975.

8. Iacoboni, M. and Zaidel, E. Hemispheric independence in word recognition: Evidence from unilateral and bilateral presentations. Brain Lang. 53, 121-140, 1996.

9. Kim, H. and Levine, S.C. Variance differences in asymmetry scores on bilateral versus unilateral tasks. Cogn. Neuropsychol. 11, 479-498, 1994.

10. Lambert, A.J. Attentional interaction in the split brain: Evidence from negative priming. Neuropsychologia 31, 313-324, 1993.

11. Levy, J., Wagner, N. and Luh, K. The previous visual field: Effects of lateralization and response accuracy on current performance. Neuropsychologia 28, 1239-1249, 1990.

12. Maljkovic, V. and Nakayama, K. Priming of pop-out: I. Role of features. Mem. Cogn. 22, 657-672, 1994.

13. Measso, G. and Zaidel, E. Effect of response programming on hemispheric differences in lexical decision. Neuropsychologia 28, 635-646, 1990.

14. Pashler, H. and Baylis, G. Procedural learning: II. Intertrial repetition effects in speeded-choice tasks. J. Exp. Psychol.: Learn. Mem. Cogn. 17, 33-48, 1991.

15. Pedhazur, E.J. Coding subjects for repeated measures designs. Psychol. Bull. 84, 298-305, 1977.

16. Rayman, J. and Zaidel, E. Rhyming and the right hemisphere. Brain Lang. 40, 89-105, 1991.

17. Snedecor, G.W. and Cochran, W.G. Statistical Methods. The Iowa State University Press, Ames, Iowa, 1976.

18. Zaidel, E. Hemispheric monitoring. In Duality and unity of the brain, D. Ottoson (Editor), pp. 247-281. MacMillan, London, 1987.

19. Zaidel, E., Clarke, J.M. and Suyenobu, B. Hemispheric independence: A paradigm case for cognitive neuroscience. In Neurobiology of Higher Cognitive Function, A.B. Scheibel and A.F. Wechsler (Editors), pp. 297-355. Guilford Press, New York, 1990.