Copyrighl (c 1988 by The Sociely for Psychophysioiogical Research. Inc. Printed in U.S.A. Toward a Functional Categorization of Slow Waves DANIEL S. RUCHKJN, Department of Physiology. School of Medicine, University of Maryland, Baltimore, Maryland RAY JOHNSON, JR., Clinical Neuropsychology Section, National Institute of Neurological. Communicative Disorders & Stroke. Medical Neurology Branch, National Institutes of Health, Bethesda, Maryland DAVID MAHAFFEY, ARD Corporation, Columbia, Maryland AND SAMUEL SUTTON Department of Psychophysiology. New York State Psychiatric Institute, New York, New York ABSTRACT This study is concerned with slowly varying, long-duration brain event-related potential (ERP) components, referred to as Slow Wave activity. Slow Wave activity can be observed in the epoch following P3b, suggesting that it reflects further processing invoked by increased task demands, beyond the processing that underlies P3b. The present experiment was designed to distinguish Slow Wave activity related to specific types of task demands which arise during difficult perceptual (pattern recognition) and conceptual (arithmetic) mental operations. Three late ERP components that respond differentially in amplitude to manipulation of perceptual and conceptual difficulty were identified: 1) A P3b, with a topography focused about Pz, evidently related to the subjective categorization of easy and difficult conceptual operations, that increased when the subjective low-prohability operation was performed; 2) A longer latency, centroparietal positive Slow Wave that increased directly with perceptual difficulty but was not affected by conceptual difficulty; 3) A very long latency negative Slow Wave, broadly distributed over centroposterior scalp, that increased directly with conceptual difficulty while its onset was delayed when perceptual difficulty increased. DESCRIPTORS: Event-related potentials. Slow Wave, P3b, Mental arithmetic. Pattern recognition. Information processing. view). These "Slow Waves" can be observed in the epoch following P3b, suggesting that they reflect further processing invoked by increased task de- mands, beyond the processing that underlies P3b. Slow Wave activity has been found in a variety of tasks. Comparisons of these data suggest that there are systematic differences in Slow Waves such that they appear to be reducible to two broad cat- egories based on their onset latencies, those reflect- jng gj^jjer (1) perceptual operations or (2) concep- Several reports have described slowly varying, long-duration brain event-related potential (ERP) components whose amplitudes relate directly to task demands (see Ruchkin & Sutton, 1983, for a re- This research was supported in part by a U.S.P.H.S. grant from the National Institute of Neurological and Communicative Disorders and Stroke, NS 11199. We are indebted to Dr. Richard Coppola for providing laboratory computer software and Mr. Howard Canoune for editorial the remaining authors plttude tncreases when the sttmulus parameters are.address requests for reprints to: D.S. Ruchkin, De- manipulated such that perception is made difficult partment of Physiology, School of Medicine, University as in signal detection, recognition, and localization of Maryland, Baltimore, Maryland 21201, USA. paradigms (Ruchkin, Sutton, Kietzman, & Silver,
340 Ruchkin, Johnson, MahafFey, and Sutton Vol. 25, No. 3 1980: Kok & de Jong, 1980; Parasuraman, Richer, & Beatty, 1982; Sutton, Ruchkin, Munson, Kietzman, & Hammer, 1982; McCallum, Curry, Cooper, Pocock, & Papakostopoulos, 1983). Slow Waves are also elicited when diflicult conceptual operations are required as in matching of linguistic stimuli (Sanquist, Rohrbaugh, Syndulko, & Lindsley, 1980), semantic judgment/incidental memorization (Neville, Kutas, Chesney, & Schmidt. 1986), complex mnemonic memorization activity (Karis, Fabiani, & Donchin, 1984), memory search (Okita, Wijers, Mulder, & Mulder, 1985; Kramer, Schneider. Fisk, & Donchin, 1986), retrieval from memory of abstract information (Rosier, Clausen, & Sojka, 1986), leaming when transitions occur in stimulus sequence-generating rules (Stuss & Picton, 1978; Stuss, Toga, Hutchison, & Picton, 1980; Johnson & Donchin, 1982), processing low-probability stimuli in a Bernoulli sequence' (N. Squires. K. Squires, & Hillyard, 1975), and mental rotation (Stuss, Saraziti, Leech. & Picton 1983; Peronnet & Farah, 1987; Johnson. Cox, & Fedio, 1987). The data from these studies indicate that there are at least three parameters of Slow Wave activity that vary as a function of task; 1) onset latency. 2) topography, and 3) polarity. Comparisons across reports in which Slow Wave onset latency can be estimated suggest that this component may be initiated at the time when the additional processing effort is required. Thus, onset latencies are relatively short (150-450 ms) for difficult perceptual operations (e.g., detection and identification of an external stimulus). Although Slow Wave onset overlaps P3b for perceptual processing, its duration is markedly longer than P3b. In contrast, for mental processes that follow perception (conceptual operations), the range of Slow Wave onset latencies increases to 300-800 ms. Slow Wave topography also varies across tasks. Slow Waves associated with perceptual difficulty are generally positive over posterior and central scalp. In some studies it was also positive, but lower in amplitude, over frontal scalp (Johnson & Donchin, 1978; Kok & de Jong, 1980), while in other studies the frontal aspect of Slow Wave was negative (McCallum et al., 1983; Ruchkin, Sutton, Kietzman. & Silver. 1980; Ruchkin, Sutton, & Stega, 1980; Sutton et al., 1982). When a feedback task 'The Slow Wave activity in Oddball tasks may be viewed as a special case of the tasks involving detection of changes in the sequence-generating rule, since their amplitudes are inversely related to event probability. Their relatively short duration may reflect the relative ease of categorizing events as having either a high or a low probwas combined with difficult signal detection. Slow Wave was larger centrally for the combined task than for signal detection alone (Johnson & Donchin, 1978; Ruchkin, Sutton, & Stega, 1980). There is an even greater variety of Slow Wave topographies associated with various conceptual operations. For example, positive Slow Waves, maximal over centro-posterior scalp, have been obtained during reactivation of stored abstract itiformation (Rosier et al., 1986), semantic matching (Sanquist et al., 1980), and rule leaming (Stuss & Picton, 1978; Stuss et al., 1980; Johnson & Donchin, 1982). A Slow Wave that was positive over posterior scalp, negligible at central scalp, and negative over frontal scalp was elicited by low-probability stimuli in a Bemoulli sequence (N. Squires et al., 1975). Karis et al. (1984) obtained a relatively large, frontal-maximal positive Slow Wave in an were employed during overt memorization, while Neville et al. (1986) obtained a Slow Wave that was large and positive over both parietal and frontal scalp in a semantic judgment/incidental memory paradigm. However, the Neville et al. experiment also involved perceptual difficulty, so their Slow Wave data may have also reflected perceptual operations. In memory scanning experiments, Okita et al. (1985) found short duration (400-700 ms) negative Slow Waves that were maximal over centro-posterior scalp, while Kramer et al. (1986) reported longer-duration (500-1200 ms) negative Slow Waves that were maximal over frontal scalp. Stuss et al. (1983) and Peronnet and Farah (1987) reported centro-parietal negative Slow Waves during mental rotation of visual images, and Stuss et al. showed that the late negative component was not present for other tasks involving complexfigures. Peronnet and Farah demonstrated that the amplitude of their negativity was directly related to the extent to which the figure had to be rotated. Finally, Johnson et al. (1987) obtained a negative Slow Wave with a frontal topography in a mental rotation experiment, although it is unclear to what extent this activity reflects true rotational processing since the stimuli could be verbally encoded. All of the above studies have the common characteristic that Slow Wave amplitude increased as a function of task demand and/or with behavioral signs of improved processing efficacy (e.g., better recall in memorization paradigms). Latency differences in these studies suggest that Slow Wave reflects when the processing occurs, while topographic differences as a function of task suggest that the Slow Waves may reflect the nature of the additional processing, in the sense that different tasks may in-
.\fay. 1988 Functional Categorization of Slow Waves 341 volve different tieural generators. With some exceptiotis, there appears to be a pattem in these results such that negative Slow Waves are associated with scanning and mental imagery while positive Slow Waves are associated with memory storage, rule learning, and perceptual operations. These differences suggest that by suitable manipulation of task type and task difficulty, it may be possible to delineate ERP reflections of specific types of sustained mental operations. The experiment reported here was designed to study the similarities and differences in Slow Wave components associated with perceptual (pattern recognition) and conceptual (mental arithmetic) operations. Our approach was to manipulate independently two levels of pattem recognition and three levels of arithmetic difficulty randomly from trial to trial. Methods Subjects Twelve right-handed volunteers (9 male) with a median age of 25 years (range 18-33) served as subjects. All subjects had normal or corrected-to-normal vision and were paid $7.00/hour. Experimental Design and Procedure Conceptual difficulty (three levels) and perceptual difficulty (two levels) were varied in a factorial manner. Conceptual difficulty was manipulated by requiring the subjects to either memorize a number (easy), use it in mental subtraction (less easy), or use it in mental division (difficult). Perceptual difficulty was manipulated by varying the discriminability of the number. It was either intact (easy) or degraded (difficult). The number was formed by a 5 X 7 matrix of dots and blanks. In the difficult perception condition, 5 or 6 dots/blanks in the matrix were randomly inverted, thereby creating blanks where the intact character had dots and inserting dots where the intact character had blanks. Subjects initiated trials by moving their right forefinger or index finger into the path of a beam of an optical sensing switch. After an 800-ms delay, the task stimulus, a visual display consisting of a row of four characters, appeared for 500 ms. The lef\-most character, a lower-case alphabetic letter, indicated the operation to be performed: "r" for remember, "s" for subtraction, and "d" for division, while the right-most three characters were numbers. For the remember operation, subjects memorized the right-most digit. For the subtraction operation, subjects computed the magnitude of the difference between the right-most and adjacent, center digit. For tbe division operation, subjects computed the remainder from the division of the right-most digit into the number composed by the other two digits. Each operation resulted in a single digit that was compared with a probe digit presented 3.2 s after the task stimulus. Subjects indicated via a choice reaction time response whether the probe digit matched or mismatched the result of their mental operation. Subjects were instructed to respond promptly, but not at the expense of accuracy, to the onset of the probe. The probe display was terminated by the subject's response or when 5 s had elapsed from probe onset. Across subjects, the median of the within subject median intertrial interval was 9.2 s. Using three independent random sequences, the 12 permutations of three levels of conceptual difficulty X two levels of perceptual difficulty X two match/mismatch conditions were equiprobable and varied randomly from trial to trial. The combinations of the letter-digit-digit-digit stimuh were determined as follows. For division, the divisor (right-most digit) was in the range of 6 to 9. The dividend (central two digits) was in the range of 21 to 99, subject to the restriction that it was not a multiple of 10 and the remainder of the division could not be zero. There are a total of 251 three-digit combinations that satisfy these conditions which were used to construct a table of 753 four-character combinations by pairing each three-digit combination with "d", "s" and "r", respectively. A set of 420 combinations were drawn from this table by random sampling without replacement, giving a total of 137 division trials, 142 subtraction trials, and 141 remember trials. After randomly subdividing for perceptual difficulty and match/mismatch outcome, there were 207 trials with all characters intact and 213 trials with a degraded character, and there were 220 mismatch and 200 match trials. All subjects were presented with the same sequence of six blocks of 70 trials. Prior to the experimental session each subject participated in a practice session of 210 trials. None of the trials in the practice set were used in the experimental session. Subjects were seated in a dimly lit, electrically shielded, sound-attenuating chamber. The stimulus was a row of white characters on a dark grey background displayed on a cathode ray tube screen (CRT) with a rapid decay phospher (P4). The CRT was located 140 cm in front of the subject, outside the shielded room, and viewed through a window. Intensity was adjusted so that characters could be clearly distinguished with minimal after-image effects (Brown. 1965). The row of characters was 2.4 cm wide X.9 cm high (approximately 1.0 X 0.4 degrees of arc) and was located slightly below eye level. A fixation light (.5 cm diameter green LED), located directly under the center of the row, was on throughout the trial. Subjects were instructed to make no eye movements during the period that started with their trial initiation finger movement and ended with probe onset. Recording Procedures ERPs were recorded with Ag/AgCl electrodes from occipital (OyJ, parietal (P^), vertex (Cz), frontal (Fz), and prefrontal (Fpz) sites. Lateral placements were at Cj and C4. Linked eariobes were used as the reference. The ground electrode was on the left wrist. All imped-
342 Ruchkin, Johnson, Mahaffey, and Sutton ances were below 1000 ohms. In order to avoid possible spurious lateral asymmetries in ERP recordings, the impedances of the two reference electrodes were equalized by means of an adjustable resistor inserted in series with the lower impedance electrode (Garnetsky & Steelman, 1958; Goff, 1974). In order to counter possible differences in electrode polarization, amplifiers, and analog-to-digital converter channels, the scalp location at which each electrode was placed and the amplifier and A-to-D channel used for a given scalp location were systematically varied across subjects. Eye movement (EOG) artifact was monitored by electrodes placed below the outer canthus and above the inner canthus of the left eye. The amplifiers were set to an upper cutoff frequency (-3dB) of 30 Hz, an AC couphng time constant of 5.3 s, with a gain of 10000 for the scalp channels and 5000 for the EOG. The recording epoch extended from 300 ms before to 2700 ms after task display onset. A digital computer controlled stimulus selection and timing, recorded response choices and reaction times (1 ms resolution), digitized (sampling rate of 100 samples/s) and monitored data on-line, and stored the data n digital format for subsequent analyses. Prior to further the r e (rms) a plitude of the EOG was computed for each trial. If the rms amplitude of the EOG exceeded 15 microvolts (MV). or if the subject moved his/her finger prior to probe onset, the trial was excluded from the analyses. Data Analysis Procedures For each subject, average ERPs were computed at all electrode sites, pooled across match and mismatch trials, for each permutation of conceptual difficulty (divide, subtract, and remember) and perceptual difficulty (degraded and intact character). The analyses reported here are confined to data from correct responsp ials. A calibrated amplifier ground was used to assess possible pre-trial shifts in voltage and to permit conversion of the average ERPs obtained with the 5.3-s time constant to waveshapes approximating those that would have been obtained by DC recording (Gaillard & Naatanen, 1980; Elbert & Rockstroh, 1980; Ruchkin, Sutton, Mahaffey, & Glaser, 1986). Since no significant differences were found among pre-trial voltage amplitudes as a function of experimental conditions, the average voltage amplitude in the 300 ms pre-task display interval was used as the baseline for amplitude measurements. Baseline-to-peak amplitudes and cross-products matrix principal components varimax analysis (PCVA) were used to obtain measures of component amplitudes. A 2400-ms time epoch beginning at task stimulus onset was used for the PCVA. Prior to PCVA computation, the input waveforms were smoothed by a digital low-pass filter with a cutoff frequency (-3dB attenuation) of 3.4 Hz. The resuhing amplitude weighting coefficients (factor scores) were converted to microvolts (Ruchkin, Sutton, Munson, Silver, & Macar, 1981). Multivariate analysis of variance (MAN- OVA) was used to assess statistical significance of the Vol. 25, No. 3 results. Since MANOVA takes into account correlation among levels that may occur in a repeated measures design, it does not understate the likelihood of Type 1 errors (Jennings & Wood, 1976; Monison, 1976; Vasey & Thayer, 1987). Profile analyses were utilized to determine whether a pair of amplitude measurements obtained at different latencies reflected the activity of more than one population of neurons (i.e., more than one component). Profiles of baseline-to-peak amplitude measurements over either electrode location (topographic comparison) or as functions of variation in experimental conditions (functional comparison) were compared. If the same neural generator underlies the ERP measurements at two different latencies, then the shapes of the pair of topographic profiles should be the same, and the shapes of the pair of functional profiles should be the same. Hence, a difference in either the topographic or functional profiles would be evidence that more than one neural generator contributes to the ERP meaparisons were confined to shapes alone, the data were first scaled so that the rms amplitudes of the means of the measurements (averaged across subjects) from the two epochs were the same (McCarthy & Wood, 1985). Thus, given shape differences, a MANOVA on the scaled data should produce a significant interaction between the independent variable (either electrode location or experimental condition) and location of measurement epoch (latency). Results Accuracy and Reaction Time Behavior Increased difficulty of both perceptual and conceptual operations reduced response accuracy and increased median reaction time to the probe presented 3200 ms after onset of the task stimulus (see Table 1). A MANOVA was formulated to test the significance of the effects of variation of perceptual and conceptual difficulty jointly upon accuracy and reaction time. The behavioral data were treated as a two-dimensional variable with accuracy and reaction time as the dimensions. Task performance was significantly affected by both perceptual difficulty (F(2/10)=62.22, ;7<.OOOO1) and conceptual difficulty (f(4/8)= 12.99, /7=.OO14), while the interaction between these main effects was not statistically significant (/7=.l 1). Overall ERP Findings A P3b and two different Slow Wave components that related differently to perceptual and conceptual operations were elicited by the task stimuli. Across subject ERP averages elicited by the task stimulus at the midline electrodes are presented in Figure 1, where the three levels of conceptual difficulty are superimposed. The effects of perceptual difficulty can be seen more clearly in Figure 2, where the two
Functional Categorization of Slow Waves ;s subject medians of percentage ei Level of Perceptual P rcentage Errors Remember Subtraction Division Reaction Times (ms) Remem Kr Subtraction Division Easy Difficult 0.0 10.2 1.4 14.0 691 704 740 719 862 Manova Manova Source df F df F P 'erceptual Difficulty (P) Conceptual Difficulty (C) P X C 1/11 2/10 2/10 135.41 20.98 3.63.00001.00026.065 1/11 2/10 2/10 13.14 1.89 3.48.004.20.071 EASY PERCEPTION DIFFICULT PERCEPTION -DIVISION - SUBTRACTION REMEMBER 0 900 1800 2700 MSEC Figure 1. ERP waveforms, averaged across subjects, for all three levels of con ptual difficulty and both levels of perceptual difficulty. Solid lines: division; dotted lines: subtraction; dashed lines: r member. In thisfigureand Figures 2 and 3, task stimulus onset and offset are indicated by the solid and dashed vert respectively. py The vertical level of the time axis corresponds to the prestimulus baseline amplitude. These waveforms were smoothed by a lowpass, zero-phase shift digital filter with a cutoff frequency (- 3dB attenuation) at 5 Hz. The EOG recording was such that vertical movements and blinks produced deflections of the same polarity in the ECXJ and scalp recorded wave- waveforms from the remember condition are superimposed. It is evident from these data that there was a large positive component with a peak latency of about 700 ms, maximal at Pz, whose amplitude was slightly lai^er for division trials in comparison with subtract and remember trials. Given its latency and topography, this component is tentatively referred to as P3b.
344 Ruchkin, Johnson, Mahaffey, and Sutton REMEMBER 900 lboo MSEC -EASY PERCEPTION DIFFICULT PERCEPTION Figure 2. ERP waveforr averaged across subjects, om midline electrodes that c ltrast ERPs from easy (sohd ne) and difficult (dotted li :) perception trials for the :onditi Following P3b, two distinct patterns of slow activity were present in the waveforms. First, as perceptual difficulty increased, there was an increased positivity over centro-parietal scalp that began at about 750-ms latency and lasted at least 800 ms (see Figure 2). Since its onset latency, topography, and relation to increased perceptual difficulty were similar to those of the component commonly referred to as Slow Wave (Ruchkin & Sutton, 1983), it will be so designated in this report. Second, a very long-duration component, lasting to at least the end of the recording epoch (2700 ms) and referred to as Late Slow Wave, was negative over centro-posterior scalp, small at F^, and positive at Fp7. The amplitude of this component increased directly with conceptual difficulty, an effect that became apparent in the 900-1200 ms latency range (see Figure 1). ERP Difference Waveforms To delineate further the effects of perceptual and conceptual difficulty, difference waveforms were obtained by using the ERP for the easy perception, remember condition as a baseline and subtracting it from the ERPs for the subtraction and division conditions (see Figure 3). Deviations from this baseline for the subtraction and division conditions characterize the effects of perceptual and conceptual difficulty upon the ERPs. Note that this difference wave formulation did not use data from the difficult perception, remember condition (dotted waveforms in Figure 2). Figure 3 indicates that P3b, Slow Wave, and Late Slow Wave components displayed different topographic, temporal, and functional characteristics in response to the changes in experimental conditions. In the P3b latency range (500-800 ms), amplitude at Pz increased for division but not subtraction, and the effect of perceptual difficulty was negligible. In the Slow Wave latency range (800-1500 ms), the positivity increased directly with perceptual difficulty over centro-posterior scalp while conceptual difficulty had a relatively small effect in this latency epoch/scalp region. The largest ERP response, which was in the Late Slow Wave latency range (1000-2700 ms), showed a marked increase in magnitude (negative over posterior scalp, positive over prefrontal scalp) with increased conceptual difficulty. This "conceptual" effect was relatively immune to perceptual difficulty during division. Baseline-to-peak amplitude measures, averaged over a component's latency range, were obtained from the ERP difference waveforms. Confounding effects of component overlap were minimized by choosing epochs that restricted the average to latency regions where the component of interest tended to dominate. Thus the P3b, Slow Wave, and Late Slow Wave were measured over 500-800 ms, 1000-1300 ms, and 2000-2700 ms, respectively. Across subject averages of these difference amplitude measures are plotted as a function of electrode position for all levels of conceptual and perceptual difficulty in Figure 4. Since analysis of multi-component, multi-electrode ERP data involves a relatively lat^e number of statistical tests, there can be a higher risk of falsely rejecting the null hypothesis than for simpler data involving relatively few tests. To deal with this possibility, a single global MANOVA was used to first test effects of perceptual and conceptual difficulty upon the difference waveform amplitude measures. The data from all seven electrodes were treated as three-dimensional vectors, with one dimension each for P3b, Slow Wave, and Late Slow Wave ampli-
May, 1988 DIVISION Functional Categorization of Slow Waves SUBTRACTION PERCERTION -EASY DIFFICULT Fpz ^ 0 900 MSEC Figure 3, Difference waveforms division ERPs (left column) and s difficult perception (dotted lines) a I I I i I I I i 0 900 1800 MSEC 'eraged across subjects. Easy percej raction ERPs (right column). Wav superimposed. nember ERPs were subtracted from Dr easy perception (solid lines) and tude. The factors of the MANOVA were perceptual difficulty (easy, difficult), conceptual difficulty (subtraction, division), and electrode location (O7, Pz, C, C4, Cz, Fz, Fpz). There were significant main effects of perceptual difficulty (f(3/9)=5.49,p=.02) and conceptual difficulty (F(3/9) = 9.08, p =.0044) with an insignificant interaction (p=.29). There were not enough degrees of freedom for testing electrode effects over seven levels. = Separate MANOVAs were then formulated to evaluate the significance of the effects of perceptual and conceptual difficulty in each component epoch. To ensure that tests involving scalp topography had adequate power (Vasey & Thayer, 1987), the electrode factor was restricted to the five midline locations: Oz, P/, Cy, Fz, Fpz. Then, for each of the epochs, separate two-factor MANOVAs were com- -The same pattern of results was obtamed for a similar global MANOVA of the original (not difference wavetorm) data. In the original data, there were three levels of conceptual difficulty (remember, subtract, divide). There were significant main effects of perceptual difficulty (f(3/ 9) = 9.64, /?=.0036) and conceptual difficulty (f (6/6) = 5.24. P=.O32) with an insignificant interaction (p =.22). puted to determine the significance of the effects of perceptual and conceptual difficulty at each of the five midline electrodes. The results of the three- and two-factor MANOVAs for each of the three components are presented in Tables 2 and 3, respectively. P3b. Figure 3 and Tables 2 and 3 indicate that there was a small but significant increase in P3b amplitude for division trials, the effect being largest and most consistent at Pz. Perceptual difficulty had a negligible effect on P3b amplitude at centro-pos- Slow Wave. Tables 2 and 3 indicate that Slow- Wave was powerfully affected by perceptual difficulty. For the subtraction condition, in which the overlapping Late Slow Wave was small, the increase in positivity with increased perceptual difficulty was restricted to the 800-1500 ms latency epoch and the Pz and Cz sites (right column of Figure 3). The interaction of conceptual difficulty with electrode may have been due to increased Late Slow Wave activity in the division condition which overlaps the Slow Wave epoch and was predominant at Oz and Fpz. The difference between the midline electrode ERPs for easy and difficult perception
Ruchkin, Johnson, Mahaffey, and Sutton Table 2 MANOVA on baseline-to-peak amplitude measures of difference ERPs P3b (500-800 ms) Slow (1000- Wave 300 ms) Late Slow Wave (2000-2700 ms) Source df Perceptual (P) :onceptual (O P X C Electrode (E) P X E P X C X E 1/11 1/11 1/11 4/8 4/8 4/8 4/8.65 11.04.0068.02.52 4.30.038 13.22.0013.67 6.81 2.84.34 5.32 8.48 4.85.48.024.022.0050.028.84 10.21.0085 3.50 5.07.025 2.22 10.21.0031.97.\V)(('. The probability of falsely n e. the null hypothesis i LATE SLOW WAVE in the remember condition (see Figure 2), which -2.0 r ^^ SLOW WAVE did not enter into the difference waveform analyses, was also significant (fl(l/l 1)= 13.69, p=.oo35). For division, the difference waves at Oz were negligible until the onset of the negative Late Slow Wave in the 1000-1200 ms latency range. This contrasts with the difference waveforms at C^, where the positive Slow Wave emerges at about 750 ms latency. The differing onset times at Oz and Cz, coupled with the observation that Slow Wave was confined to Cz and Pz in the subtraction condition, are consistent with an interpretation that the ERP activity at Oz in the 1000-1300 ms epoch is due to Late Slow Wave and not to Slow Wave. At Fpz, the amplitude profile differed from the respective profiles at Cz and Pz as a function of conceptual and perceptual difficulty. Whereas the effect of conceptual difficulty was relatively large at Fpz, the effect of perceptual difficulty was relatively small (see Figure 3). This pattem of variation was opposite to that in the Slow Wave epoch at Cz and Pz but was the same as that in the Late Slow Wave epoch. A MANOVA comparison of the profiles at Fpz and Cz revealed significant differences across experimental conditions (F(3/9)= 10.06, p=.oo18), indicating that two or more components contributed to the Slow Wave measurements at prefrontal and centro-parietal scalp. This finding supports an interpretation that ERP variations in activity at Fpz in the 1000-1300 ms epoch were due primarily to overlap from the prefrontal aspect of Late Slow DIVISION, EASY PERCEPTION Wave, -- DIVISION. DIFFICULT PERCEPTION SUBTRACTION. EASY PERCEPTION Late Slow Wave. Figure 4 and Tables 2 and 3 SUBTRACTION, DIFFICULT PERCEPTION clearly indicate that Late Slow Wave was powerfully affected by conceptual difficulty. There were Figure 4. Mean baseline-to-peak amplitudes, averaged relatively large, consistent increases in posterior across subjects, of the ERP difference waveforms in the Late Slow Wave epoch (top panel). Slow Wave epoch scalp negativity and prefrontal positivity as conceptual difficulty increased from subtraction to di- (center panel), and P3b epoch (bottom panel) plotted as functions of midline electrode, conceptual task, and level vision. In contrast, perceptual difficulty produced of perceptual difficulty (easy, difficult). a small, statistically weak effect that was opposite
Functional Categorization of Slow Waves Table 3 Significant main effects of perceptual and conceptual difficulty upon baselinepeak amplitude measures of difference ERPs at each midline electrode P3b (SOO-800 ms) Slow Wa' (1000-1300 Late Slow Wave Perceptual p, CV +0.91 + 1.80 + 1.67.0059.OOO66'» Fiv -0.85.0075-1.19.046-1.16.033 Conceptual 0/ C, + 1.10 + 0.41.0085.038-0.96.023-3.25.00001*" -3.96.00001"* -2.24.0015* + 1.56.011 + 3.06.0018 + 3.69.0023 M'. The Amplitude columns contain the amplitude difference (difficult minus easy.ion) as the level of difficulty increased in each case, pooled across the easy and difficult ions for the other case; e.g.. the amplitude differences for the conceptual case are pooled the easy and difficult perception conditions. <.O5 (.05/30), **p<.02 (.02/30), ***p<.01 (.01/30). by the Conservative Bonferroni to the effect for increased conceptual difficulty. As perceptual difficulty increased, the posterior negativity and the prefrontal positivity in the division condition tended to decrease. Activity in the 1000-1300 ms epoch at Oz also suggests that the onset of Late Slow Wave was later as perceptual difficulty increased (left column of Figure 3). ERP Actiyity at F,./. The data in Figure 3 and Table 3 suggest that the ERP activity at Fpz primarily reflected the activity of one long-duration component that overlapped the P3b, Slow Wave, and Late Slow Wave measurement epochs. As noted above, the profile of Fpz variation in the Slow Wave epoch differed from the centro-parietal profile when perceptual and conceptual difficulty were varied. Increases in perceptual difficulty also had different effects upon the ERPs elicited at Fpz and Pz in the P3b epoch: amplitude at Fpz was reduced while there was negligible change in amplitude at P/. ERP activity at Fpz in all three epochs most closely resembled that of the posterior Late Slow Wave except that its polarity was reversed. Thus the ERP activity at Fpz could be a reflection of the positive pole of the neural generator that underlies the negative Late Slow Wave activity observed over posterior scalp. There were, however, some differetices between the time courses of the prefrontal and posterior aspects of these ERPs. For example, the Oz difference waveforms suggest that the onset of Late Slow Wave occurred in the 1000-1200 ms latency range whereas amplitude increased in the 600-800 ms latency range at Fpz. Comparison of the results of the Oz and Fpz tests of the ERP data in the 500-800 ms epoch indicates that conceptual and perceptual difficulty produced significant effects at Fpz but not at Oz (Table 3). Consequently, it is possible that the Fpz data reflect a component that, in some respects, is similar to the negative, posterior Late Slow Wave, but with an earlier onset. Difference Wayes PCVA. The above analyses indicate that P3b, Slow Wave, and Late Slow Wave are different, partially overlapping components. A PCVA of the difference ERPs was implemented in order to obtain approximations of component waveshapes and amplitudes that may be relatively free of temporal and spatial overlap. The input data were 336 difference waveforms (2 levels of perceptual difficulty X 2 levels of conceptual difficulty X 7 scalp electrodes X 12 subjects). Three components, spanning 87% of the energy of the data set, were extracted. The component basis waveforms, viewed as estimates of P3b, Slow Wave, and Late Slow Wave, are shown in Figure 5. These basis waveforms indicate that the components occurred in a partially overlapping sequential temporal order. Approximate onset, peak, and offset latencies for the high-enet^y portions of the basis waves were as follows: P3b-3OO ms, 750 ms, 1400 ms; Slow
LATE SLOW WAVE ^SLOW WAVE -P3b Figure 5. Basis waveforms obtained from the crossproducts PCVA of the ERP difference waveforms. The polarities have been adjusted so that each basis wave reflects the polarity (positive down, negative up) of the com- Wave-500 ms, 1100 ms. 2000 ms: Late Slow Wave-1000 ms, greater than 2400 ms, greater than 2400 ms.' Profile Comparisons Functional Comparisons. The issue of whether different neural generators underlie P3b, Slow Wave, and Late Slow Wave in the difference waveforms was examined by pairwise functional profile comparisons (pooled across P^ and Cz) over the four {Permutations of perceptual difficulty with conceptual difficulty. Comparisons were made for inverted Late Slow Wave (i.e., same polarity as P3b and Slow Wave) and non-inverted Late Slow Wave. For the comparisons of Slow Wave versus P3b and Slow Wave versus inverted and non-inverted Late Slow Wave, the MANOVA levels of significance were all less than p =.0074, indicating that the functional profile of Slow Wave at C^ and?z clearly differed from those for P3b and Late Slow Wave. Hence, the Slow Wave neural generator may differ from those of Late Slow Wave and P3b. For comparisons 'The pattern of variation of the PCVA weighting coefficients as functions of electrode position and levels of conceptual and perceptual difficulty was very similar to the pattern of variation of the baseline-to-peak amplitudes. A global MANOVA on the weighting coefficients revealed significant main effects for both perceptual (F(3/ 9)=3.85, p=.q5) and conceptual difficulty (F(3/9)=6.OO, p=.o16). with no significant interaction between the two (p=.62). The patterns of results of the three-factor and Ruchkin, Johnson, MahafFey, and Sutton Vol. 25. No. 3 between P3b and the Late Slow Wave, the significance levels were /7=.O13 and p=.53 for the inverted and non-inverted Late Slow Waves, respectively, providing some support (the non-inverted Late Slow Wave comparison) for a conclusion that P3b and Late Slow Wave have separate underlying neural generators. Topographic Comparisons. The issue of whether the observed P3b, Slow Wave, and Late Slow Wave components were generated by different populations of neurons was further examined by means of topographic profile comparisons on the original, unsubtracted ERP waveforms. Pairwise comparisons between components were made on baselineto-peak amplitude measurements from Oz, Pz, and Cz, pooled over all experimental conditions. Comparisons were restricted to the electrodes where the components were largest to reduce the likelihood of contamination by other components and to increase the statistical power of the MANOVA. As above, comparisons were made with inverted and non-inverted Late Slow Waves. Late Slow Wave topography clearly differed from those of P3b and Slow Wave. For each of these four tests, the level of significance was less than p=.0087. P3b and Slow Wave topographies were also found to be significantly different (p=.o18). An implicit assumption of the baseline-to-peak and PCVA analyses is that these measurements are minimally affected by latency variability as a function of condition. Nevertheless, it is possible that the relation between Slow Wave amplitude and perceptual difficulty was due to either: 1) the later onset of Late Slow Wave for increased perceptual difficulty "unmasking" a constant positivity, or 2) increased duration of P3b as perceptual difficulty increased. These hypotheses were examined with profile comparisons of Slow Wave amplitudes over Oz, Pz, and Cz after subtracting the easy perception condition from the difficult perception condition and pooling across levels of conceptual difficulty. If the increased positivity of Slow Wave was due to the later onset of the negative Late Slow Wave, then the topography of the Slow Wave difference measure should be similar to that of the inverted Late Slow Wave. Two measures of Late Slow Wave were used: 1) average amplitude over the 2000-2700 ms epoch (the measure used in all other analyses), and 2) average amplitude over the 1500-2000 ms epoch (the epoch adjacent to the Slow Wave epoch). Profile comparisons were made between the inverted Late Slow Wave obtained for division trials, pooled across levels of perceptual difficulty, and the Slow Wave easy minus difficult perception difference measure. The results of the two MANOVAs, /7=.OO31 for the 2000-2700 ms epoch andp=.0014
May. 1988 Functional Categorization of Slow Wa\ 349 for the 1500-2000 ms epoch, demonstrated that the relation between Slow Wave amplitude and perceptual difficulty is not likely to be due to the unmasking of a constant positivity by a delayed onset of the negative Late Slow Wave. If the increased positivity of Slow Wave resulted from an increased duration of P3b, then the easy versus difficult perception difference measure of Slow Wave topography should be similar to the topography of P3b. Therefore, a profile comparison was made between the amplitude of P3b, pooled across all conditions, and the Slow Wave difference measure (p=.0s2). To increase the power of the MANOVA, the comparison was then restricted to the Pz and Cz electrodes, where P3b and Slow Wave were largest (p=.o3o). These results provide weak support against the hypothesis that the direct relation between Slow Wave activity and perceptual difficulty is due to an increase in P3b duration. Profile Comparisons Summary. All functional and topographic profile comparisons strongly support the conclusion that different neural generators underlie Late Slow Wave and Slow Wave. Late Slow Wave and P3b topographic profiles were also clearly different. The topographic comparison of the Slow Wave easy versus difficult perception difference measure and P3b profiles provides a weak rejection of the possibility that the increased Slow Wave activity that accompanied increased perceptual difficulty is due to increased duration of P3b. The functional comparison of Slow Wave and P3b provides stronger support for an interpretation that they arise from different neural generators, since Slow Wave amplitude clearly increased with increased perceptual difficulty and P3b amplitude did not change, while P3b amplitude significantly increased with conceptual difficulty and Slow Wave did not. Lateralization Findings Lateralization of ERP components was examined by comparing data from the C, and d electrodes. There were separate analyses for the original waveforms and the difference waveforms. Global MANOVAs were formulated by treating the data as three dimensional vectors, with the dimensions betng the amplitudes of P3b, Slow Wave, and Late Slow Wave respectively. The factors were Perceptual Difficulty, Conceptual Difficulty, and Electrode Location (C,,,). The Global MANOVA of the baseline-to-peak amplitudes from the original waveforms revealed a significant main effect of electrode (F(3/9) = 4.75, p=.030) with no significant interactions. Lateralization effects were present in the division trials with significantly larger Late Slow Waves and P3bs (/7=.O27 and p=.o4o, respectively) over the left hemisphere. Across subjects, the average amplitudes in the division trials, pooled across perceptual difficulty, for Late Slow Wave were -4.0 /xv at C, and - 2.1 fiv at C4. Corresponding amplitudes for P3b were 6.4 ^V at C, and 5.5 ^V at C4. There was no significant lateralization of Slow Wave activity. For the difference waveforms, there were no significant electrode effects for any components in either the baseline-to-peak or PCVA global MAN- OVAs. Discussion Three long-latency, slow components that respond differentially to the manipulations of conceptual and perceptual difficulty used in this study have been identified: P3b, Slow Wave, and Late Slow Wave. P3b, with a topography focused about Pz, increased in amplitude in division trials. Slow Wave, with a Cz-Pz topography, increased in amplitude directly with perceptual difficulty but was not affected by conceptual difficulty. Late Slow Wave, broadly distributed over centro-posterior scalp, increased in amplitude directly with conceptual difficulty while its onset was delayed and its amplitude decreased slightly when perceptual difficulty increased. These various results are probably due to the components' activity being related to different aspects of the display. Given that the experimental conditions varied randomly from trial-to-trial and that there were no significant effects of experimental condition upon the prestimulus ERP activity, the observed behavior of these three components cannot be attributed readily to differences in anticipatory or preparatory processes. It is also unlikely that the observed behavior was due to a motoric readiness potential associated with the subject's reaction time resjjonse. The potential contribution of a readiness potential was probably small, since the onset of the probe stimulus was 500 ms after the end of the ERP recording epoch, and mean reaction times were 600-900 ms after probe onset. Thus only the portion of the readiness potential activity in the 3300-1100 ms pre-response epoch could have possibly overlapped the three components. The three components appear to index three successive steps in processing the task stimulus display. The first step, reflected by P3b, is apparently related to recognition of the type of task and possibly to an adjustment of set and/or resources necessary for performing the task. Initial processing of the left letter must have been completed for task recognition to occur. The second step, reflected by Slow- Wave, involves perceptual processing of the right-
350 most digit when that portion of the display is degraded. The third step, reflected by Late Slow Wave, involves conceptual processing which is later than the processing indexed by P3b and Slow Wave. Late Slow Wave's extended time course suggests that it indexes sustained mental operations while its amplitude appears to reflect the difficulty associated with these operations. P3b and Slow Wave The larger P3bs in the division trials may not be associated with conceptual difficulty per se, but may be the result of subjects having categorized the three equiprobable conceptual operations into only two classes: 1) easy (remember, subtraction) and 2) difficult (division). In this scheme, difficult trials would occur only half as often as the easy trials. Since P3b amplitude is inversely related to event probability, P3b would be larger for the "low-probability," difficult task trials. In a pilot experiment with only two levels of conceptual difficulty (remember and division), the P3bs were approximately equal in size. P3b amplitude was not affected by perceptual difficulty. This latter finding was unexpected, since increased perceptual difficulty is known to increase the subject's degree of equivocation about having correctly categorized the stimulus which should result in a reduced amplitude, longer latency P3b (Hillyard, Squires, Bauer, & Lindsay, 1971; K. Squires, Squires, & Hillyard, 1975; Johnson, in press; Johnson & Donchin, 1978, 1985; Kok & de Jong, 1980; Ruchkin & Sutton, 1978; Ruchkin, Sutton, Kietzman, & Silver, 1980; Ruchkin, Sutton, & Mahaffey, 1987; Fitzgerald & Picton, 1983). Thus while the behavioral data (error rate and reaction time) indicate that equivocation indeed varied, our P3b findings were not consistent with previous reports. This seemingly anomalous result can be understood when one takes into account the fact that, in contrast to past studies, we used a multifaceted stimulus. Consequently, the "stimulus" could be divided into two, or even three or four different stimuli. That is, the left-most letter designated the task, the right-most character was the essential number to be used in all tasks, and use of the remaining numbers in the middle depended upon the task. The finding that P3b amplitude was increased in division trials coupled with the findings that (1) the behavioral data were consistent with increased equivocation when perceptual difficulty was increased, and (2) there was an absence of either reduced P3b amplitudes or increased P3b latencies, strongly suggest that the P3b in our waveforms reflects processing of only the left-most, "task-designating" character. Ruchkin, Johnson, MahafFey, and Suttoi Vol. 25, No. 3 Moreover, the behavior of the subsequent Slow Wave is consistent with the idea that it reflects processing of the right-most digit, since its amplitude increased as a function of difficulty in recognizing the right-most digit. The finding that Slow Wave amplitude was directly related to perceptual difficulty and that its topography had a centro-parietal focus is consistent with the earlier studies reviewed above. However, the onset latency (500 ms for the PCVA basis wave) was longer than observed in the earlier studies (150-400 ms). The longer latency in our study may have been due to the prior processing of information delivered by other characters in the stimulus display. Thus it would appear from the P3b and Slow Wave activity that the multifaceted nature of the stimuli caused our subjects to digest the stimulus piece by piece. Late Slow Wave The relatively large amplitude of Late Slow Wave in the division task in comparison with subtraction and remember tasks suggests that it reflects mental activity involved in (a) execution of arithmetic operations and/or (b) the amount of information maintained and accessed in short-term memory (three, two, or one digits for division, subtraction, or remember, respectively). In contrast to the P3b results, it is unlikely that the increased amplitude of the Late Slow Wave in the division task is a result of subjects categorizing the three tasks into only two classes: 1) a frequent, easy class (remember, subtraction) and 2) an infrequent, difficult class (division). In a two-subject pilot study that had only two equiprobable levels of conceptual difficulty (remember and divide), the large increase in Late Slow Wave magnitude in division trials was observed for both subjects. The effect of increasing perceptual difficulty upon Late Slow Wave contrasts with the effect of increasing conceptual difficulty: while amplitude tended to be decreased by the former, it was clearly increased by the latter. These opposing effects are consistent with an interpretation that the mental processes underlying the Late Slow Wave reflect conceptual processing rather than perceptual processing of the rightmost digit. The variations in Late Slow Wave as perceptual difficulty was manipulated are probably indirect effects. The latency and reaction time data suggest that the conceptual processing reflected by Late Slow Wave is delayed by the perceptual processing indexed by Slow Wave. The diminished amplitude and increased error rates suggest that difficulty in recognizing the right-most digit interferes with the conceptual processing by increasing the subject's degree of equivocation.
May. 1988 Functional Categorization of Slow Waves Conceptual difficulty was confounded with other factors that have been traditionally labelled as "perceptual," since, as the computation became more difficult, more of the task stimulus display had to be processed. It is important to note, however, that if increased perceptual load was responsible for the Late Slow Wave, the load associated with increased difficulty in recognizing the right-most digit had different effects upon Late Slow Wave and Slow Wave. Functional profile analyses indicate that the frontal, positive aspect of Late Slow Wave differed from the positive Slow Wave and P3b observed over centro-parietal scalp. An unresolved question is whether the frontal positive/posterior negative Late Slow Wave reflects the two poles of the neural generator of a single component or two different components. In support of the single-component hypothesis, conceptual and perceptual difficulty had similar effects upon prefrontal and posterior aspects of Late Slow Wave. However, some of the data support a two-component hypothesis, such as the finding that prefrontal activity began earlier than the posterior aspect. The two-factor MANOVAs for the 500-800 ms epoch data at Oz and Fpz suggest that the apparent onset differences may be a systematic effect, since conceptual and perceptual difficulty had significant effects at Fpz but not at Oz. On balance, the apparent earlier onset of Late Slow Wave activity at Fpz favors a two-component interpretation, but with the reservation that statistical support for the earlier onset is not strong. There are functional and topographic similarities between our Late Slow Wave and some other slow negativities reported previously. It is likely that there was some form of preparation for and/or anticipation of the probe. With respect to anticipation of the probe, the posterior aspect of Late Slow Wave topography resembles that of the contingent negative variation (CNV) non-motoric E-Wave reported by Ruchkin et al. (1986). It may well be that an E- Wave was also present in the current data and that conceivably it accounts for most of the negativity observed in the remember and subtract conditions. In the current experiment, performance of the task required that information be stored and then operated upon in short-term memory. The Stuss et al. (1983) and Peronnet and Farah (1987) mental rotation experiments also involved sustained operations upon information stored in short-term memory. Peronnet and Farah obtained a negative shift over posterior scalp in the 400-800 ms latency epoch whose amplitude was directly proportional to the angle of rotation. Its timing was such that this Slow Wave could index the rotation operation per se. While the shift was largest at P7, no information was available concerning its frontal topog- 351 raphy. Stuss et al. found a sustained negativity with a later onset (700 ms) and Peronnet and Farah suggested that this negativity possibly reflected processing that followed rotation, such as "visual comparison and... discrimination." It was maximal at Pz and broadly distributed over centro-parietal scalp. The Stuss et al. negativity appeared to differ from our Late Slow Wave in that its extension over frontal scalp was larger. Moreover, in contrast to Stuss et al.'s finding of laterally symmetric activity, the amplitude of our Late Slow Wave was somewhat larger over the left hemisphere. However, in view of the differences in procedures and electrode montages employed among these two mental rotation experiments and our study, it cannot be concluded firmly that these various negative components have different scalp topographies. It is possible that they all reflect a process that is common to mental arithmetic and various mental imagery operations. A late negativity elicited by target stimuli in a memory scanning experiment (Okita et al., 1985, Figures 3 and 7) also bears functional and some topographic resemblance to our Late Slow Wave. Okita et al.'s negativity was relatively small for memory loads of one or two items and was lai^e when the load consisted of four items. Although its onset was earlier (400 ms) and duration shorter (300 ms) in comparison with our Late Slow Wave, subjects may have performed the memory scanning task more rapidly than the division task in our study. Okita et al.'s negativity was maximal at Pz and broadly distributed over posterior and frontal scalp, with the frontal aspect appearing to be more pronounced than that of both our Late Slow Wave and Stuss et al.'s mental rotation negativity. A late positive Slow Wave observed in a study of ERPs and memory (Karis et al., 1984) bears some resemblance to the positive, frontal aspect of our Late Slow Wave. The Karis et al. positivity was largest over frontal scalp, began at 600 ms, and extended to the end of the 1200-ms recording epoch. For subjects employing complex mnemonic memorization strategies, the frontal positivity was large when elicited by stimuli that were subsequently recalled and smaller when elicited by stimuh that were not subsequently recalled. When subjects employed simple, rote memorization strategies, the frontal positivity was small. The behavior of the negative and positive slow components reported by Okita et al. (1985) and Karis et al. (1984) and the apparent earlier onset of the prefrontal, positive aspect of our Late Slow Wave suggest that the prefrontal Late Slow Wave may relate to storage and retention of information in short-term memory while the posterior, native aspect may relate to utilization of information in
352 Ruchkin, Johnson, Mahaffey, and Sutton Vol. 25. No. 3 short-term memory. Differences in topographies may be due to both the parietal negativity and frontal positivity being simultaneously present in our data with no such simultaneous presence in the Okita et al. and Karis et al. studies. Conclusion The data from the current experiment, along with the results of earlier studies, suggest that Slow Wave activity may be useful for understanding how humans process external and internal events. It appears that Slow Wave activity belongs to two broad categories, perceptual and conceptual, that are separable on the basis of onset latency. Whereas the initial studies of Slow Wave activity focused upon components that were positive over posterior scalp and negative over frontal scalp, additional variations as a function of type of processing have been found along the dimensions of topography and polarity. At present there is little information available concerning the Slow Wave duration dimension. Although it may be a more difficult property to measure, the potential contribution of Slow Wave duration measurements to our understanding of the duration of mental processes makes the solution of this measurement problem worthy of attention. However, full utilization of temporal information will require a better understanding of the similarities and differences among Slow Wave components in terms of their functional roles. Such increased understanding of functional roles may also make possible further subdivision of the conceptual Slow Wave category on the basis of onset latency if a task requires successive stages of conceptual processing. The posterior negative Slow Wave activity is a recent finding. In contrast with the posterior posi- tivities, these negativities thus far appear to be associated only with conceptual operations. At present. Slow Wave topographic data raise more questions than they answer. It may well be that very different neural generators underlie posterior-positive and posterior-negative Slow Wave activity, and therefore they may constitute two distinct "species" of Slow Wave activity. However, currently the meaning of within "species" activity is unknown. For example, do the posterior negativities associated with memory scanning, mental rotation, and arithmetic have the same or different functional roles? Are the members of each species closely related, or, upon closer examination, will it turn out that the individual Slow Waves are more different than similar? Can these components be attributed mainly to general arousal processes that coincide with difficult mental operations, or are there grounds for interpreting them as more specific reflections of different cognitive operations? In summary, we have demonstrated that Slow Wave activity can be divided into two general, independently manipulable categories: perceptual and conceptual. A further subdivision of the conceptual category may also be possible. Our data show that Slow Wave activity in the perceptual and conceptual categories responds differentially to experimental variables. Given the lack of a large body of Slow Wave data, this scheme will no doubt require modification as more data become available. Nevertheless, the advancement of this structure is a tentative attempt to understand the functional significance of this kind of ERP activity. If these ideas are supported by further research. Slow Wave activity could be a useful addition to our armamentarium of ERP components for studying human information processing. REFERENCES Brown. J.L. (1965). Afterimages. In C.H. Graham (Ed.). Vision and visual perception (pp. 479-503). New York; Wiley. Gaillard' A.W.K.. & Naatanen. R. (1980). Some baseline effects on the CNV. Biological Psychology. 10. 31-39. Elbert, T.. & Rockstroh, B. (1980). Some remarks on the development of a standardized time constant. Psychophysiology. 17. 504-505. Fitzgerald. RG.. & Picton, T.W. (1983). Event-related potentials recorded during the discrimination of improbable stimuli. Biological Psychology, 17, 241-276. Garnetsky. T.M.. & Steelman. H.F. (1958). Equalizing ear reference resistance in monopolar recording to eliminate artifactual temporal lobe asymmetr>. Electroencephalographv & Clinical Neurophvsiology. 10, 736-738. Goff. W.R. (1974). Human average evoked potentials: Procedures for stimulating and recording. In R.F. Thompson & M.M. Patterson (Eds.). Biolectric recording technicfues: Part B. Electroencephalographv and human brain potentials (pp. 102-156). New York: Academic Press. Hillyard, S.A.. Squires, K.C.. Bauer, J.W., & Lindsey, P.H. (1971). Evoked potential correlates of auditory signal detection. Science. 172. 1357-1360. Jennings, J.R., & Wood. C.C. (1976). The epsilon-adjustment procedure for repeated-measures analyses of variance. Psychophvsiology. 13, 277-278. Johnson, R., Jr. (in press). The amplitude of the P300 component of the event-related potential: Review and synthesis. In P.K..Ackles, J.R. Jennings, & M.G.H. Coles (Eds.). Advances m psychophysiology (Vol. 111). Greenwich, CT: JAI Press. Johnson, R.. Jr., Cox. C, & Fedio, P. (1987). Event-related potential evidence for individual differences in a mental rotation task. In R. Johnson, Jr., J.W. Rohrbaugh,
May. 1988 Functional Categorization of Slow Wave & R. Parasuraman (Eds.), Current trends in event-related potential research. Electroencephalography & Clinical Neurophysiology, Suppl. 40, 191-197. Johnson, R., Jr., & Donchin, E. (1978). On how P300 amplitude varies with the utility of the eliciting stimuli. Electroencephalography & Clinical Neurophysiology, 44. 424-437. Johnson, R., Jr., & Donchin. E. (1982). Sequential expectancies and decision making in a changing environment: An electrophysiological approach. Psychophysiology. 19, 183-200. Johnson, R.. Jr.. & Donchin, E. (1985). Second thoughts: Multiple P3OOs elicited by a single stimulus. Psychophvsiology. 22. 182-194. Karis, D., Fabiani, M.. & Donchin, E. (1984). "P300" and memory: Individual differences in the von Restorffeffect. Cognitive Psychology. 16, 177-216. Kok. A.. & Looren De Jong, H. (1980). Components of the event-related potential following degraded and undegraded stimuli. Biological Psychology. II, 117-133. Kramer. A.. Schneider, W., Fisk, A., & Donchin. E. (1986). The effects of practice and task structure on components of the event-related potential. Psychophysiology. 23, 33-47. McCallum, W.C, Curry, S.H.. Cooper, R., Pocock. P.V.. & Papakostopoulos, D. (1983). Brain event-related potentials as indicators of early selective processes in auditory target localization. Psychophysiology. 20, l-p. McCarthy, G., & Wood, C.C. (1985). Scalp distributions of event-related potentials: An ambiguity associated with analysis of variance models. Electroencephalography & Clinical Neurophysiology. 62, 203-208. Morrison, R.F. (1976). Multivariate statistical methods. (2nd ed.) New York: McGraw Hill. Neville. H.J., Kutas. M., Chesney, G., & Schmidt, A.L. (1986). Event-related brain potentials during initial encoding and recognition memory of congruous and incongruous words. Journal of Mernorv and Language, 25, 75-92. Okita, T., Wijers, A.A., Mulder, G., & Mulder, L.J.M. (1985). Memory search and visual spatial attention: An event-related brain potential analysis. Acta Psychologica. 60, 263-292. Parasuraman, R., Richer, F., & Beatty, J. (1982). Detection and recognition: Concurrent processes in perception. Perception & Psvchophvsics, 31, 1-12. Peronnet, F., & Farah, M.J. (1987). Mental rotation: An ERP study with a validated mental rotation task. In M. Kutas & B. Renault (Eds.), Proceedings of the 4th International Conference on Cognitive Neuroscience (pp. 49-52). Paris: Dourdan. Rosier, F., Clausen, G., & Sojka, B. (1986). The doublepriming paradigm: A tool for analyzing the functional significance of endogenous event-related brain potentials. Biological Psychology. 22. 239-268. Ruchkin, D.S., & Sutton, S. (1978). Equivocation and P300 amplitude. In D. Otto (Ed.). Multidisciplinary perspectives in event-related potential research (pp. 175-177). Washington, DC: U.S. It Printing Office. Ruchkin, D.S., & Sutton, S. (1983). Positive slow wave and P300: Association and disassociation. In A.K.W. Gaillard & W. Ritter (Eds.), Tutorials in ERP research: Endogenous components (pp. 233-250). Amsterdam: North Holland. Ruchkin, D.S., Sutton, S., Kietzman, M.L., & Silver, K. (1980a). Slow wave and P300 in signal detection. Electroencephalography & Clinical Neurophysiology. 50, 35-47. Ruchkin, D.S., Sutton, S., & Mahaffey, D. (1987). Functional differences between members of the P300 complex: P3e and P3b. Psychophysiology, 24, 87-103. Ruchkin, D.S., Sutton, S., Mahaffey, D., & Glaser, J. (1986). Terminal CNV in the absence of motor response. Etectroencephalographv & Clinical Neurophysiology. 63, 445-463. Ruchkin, D.S.. Sutton, S., Munson, R., Silver, K., & Macar, F. (1981). P300 and feedback provided by absence of stimulus. Psychophysiology. 18, 271-282. Ruchkin, D.S., Sutton, S.. & Stega, M. (1980b). Emitted P300 and slow wave event-related potentials in guessing and detection tasks. Electroencephalography & Clinical Neurophvsiology. 49, 1-14. Sanquist, T.F., Rohrbaugh, J.W., Syndulko, K., & Lindsley, D.B. (1980). Electrocortical signs of levels of processing: Perceptual analysis and recognition memory. Psvchophysiology, 17, 568-576. Squires, K.C.. Squires, N.K., & Hillyard, S.A. (1975). Decision-related cortical potentials during an auditory signal detection task with cued observation intervals. Journal of Experimental Psychology: Human Perception & Performance. 1, 268-279. Squires, N.K.! Squires, K.C., & Hillyard, S.A. (1975). Two varieties of long-latency positive waves evoked by unpredictable auditory stimuli in man. Electroencephalography & Clinical Neurophvsiology, 38. 387-401. Stuss, D.T., & Picton, T.W. (1978). Neurophysiological correlates of human concept formation. Behaviora! Bi~ ology. 23, 135-162. Stuss, D.T., Toga, A., Hutchison, J., & Picton, T.W. (1980). Feedback evoked potentials during an auditory concept formation task. Motivation, motor and sensory processes of the brain. Progress in Brain Research (Vol. 54, pp. 403-409). Amsterdam: Elsevier/North Holland. Stuss, D.T., Sarazin, F.F., Leech, E.E., & Picton, T.W. (1983). Event-related potentials during naming and mental rotation. Electroencephalography & Clinical Neurophysiology. 56, 133-146. Sutton, S.. Ruchkin, D.S., Munson, R.. Kietzman, M.L.. & Hammer, M. (1982). Event-related potentials in a two-interval forced choice detection task. Perception & Psychophysics, 32, 360-374. Vasey, M. W., & Thayer, J.F. (1987). The continuing problem of false positives in repeated measures ANOVA in psychophysiology: A multivariate solution. Psychophysiology, 24, 479-486. (Mat rript r eived August 5, 1987: accepted for public y4, 1988)