International Journal of Psychophysioloa, 10 (1990) 163-170

Elsevier 163

PSYCHO 00308

Probability and inter-stimulus interval effects on the P300 from auditory stimuli

John Polich Departmen: of Neuropharmacolo~, Research Institute of Scripps Clmic La Jollu, CA 92037 (U.S.A.)

(Accepted 5 April 1990)

Kqy words: P300 (P3); Probability; Inter-stimulus interval

The relationship between target stimulus probability and inter-stimulus interval (ISI) on the P300 (P3) component of the event-related potential was assessed in two experiments. An auditory discrimination paradigm was employed wherein subjects

indicated with a finger tap response the occurrence of a randomly presented 2000 Hz target tone embedded in a series of 1000 Hz

tones. Experiment 1 presented stimuli with target probabilities of 0.10, 0.30, and 0.50 at ISIS of 1.5 or 3.0 s and produced P3

amplitudes which decreased with increases in target probability and were smaller at the shorter compared to the longer ISI.

Experiment 2 presented stimuli with target probabilities of either 0.20, 0.50. or 0.80 at ISIS of 4.0 and 10.0 s and produced P3

amplitudes which were unaffected by either variable. P3 latency demonstrated relatively few changes in either experiment. When

taken together with previous findings, the results suggest that inter-stimulus interval affects P3 amplitude by determining the amount

of processing resources available during ERP generation

INTRODUCTION

The probability with which a stimulus occurs influences the event-related brain potential (ERP), since the amplitude of the P300 or P3 component is determined by the probability of the target stimulus. The fundamental relationship between stimulus probability and the size of the P3 poten- tial can be stated straightforwardly: the less fre- quently the stimulus item is presented, the larger the component amplitude and vice versa (Sutton et al., 1965; Tueting et al., 1971; Duncan-Johnson and Donchin, 1977). This basic finding has served as the cornerstone for the theoretical interpreta- tion of P3 amplitude as reflecting the processing associated with the updating of short-term mem-

Correspondence 10: J. Polich, Dept. of Neuropharmacology,

Scripps Clinic and Research Foundation, 10666 N. Torrey Pines Road, La Jolla, CA 92037, U.S.A.

ory when target stimulus information is acquired (Squires et al., 1976; Sutton, 1979; Donchin. 1981; Klein et al., 1984; Johnson, 1986; Donchin and Coles, 1988) and has been exploited fruitfully in the assessment of normal memory function (Polich et al., 1983; Karis et al., 1984; Fabiani et al., 1986; Neville et al., 1986; Paller et al., 1988).

Although target stimulus probability is a criti- cal determinant of P3 amplitude, the time between stimulus events or the inter-stimulus interval (ISI) also has been fqund to affect P3 magnitude. Several studies have reported that P3 components elicited with relatively short ISIS are smaller in amplitude than those obtained with longer ISIS (Woods et al., 1980; Fitzgerald and Picton, 1981: Woods and Courchesne, 1986). These effects sometimes have been attributed to ‘temporal probability’, since P3 amplitude appears to be influenced by the temporal frequency with which a target stimulus occurs in a given time interval (Picton and Stuss, 1980). Indeed. increases in P3

0167-8760/90/$03.50 19 1990 Elsevier Science Publishers B.V. (Biomedical Division)

164

amplitude with decreases in the temporal frequen- cy of the target stimulus have been observed for both easy and hard stimulus discrimination tasks in an additive fashion, suggesting that task diffi- culty manipulations may arise from temporal pre- sentation effects (Fitzgerald and Picton, 1984; Polich. 1987a). Thus, target stimulus probability and the temporal parameters of the target stimu- lus both contribute to the size of the P3 ERP.

One possible explanation for the influence of IS.1 on P3 amplitude is suggested by the similar declines in amplitude observed with decreases in ISI for sensory evoked potentials (e.g. Davis et al., 1966; Roth et al., 1976; Polich et al., 1988). These effects have been interpreted as the result of ‘re- covery cycle’ limitations inherent in the mecha- nisms responsible for component generation. Rel- atively small potentials will be produced with short ISIS because the system requires time to recover from very recent evoked potential production. With longer ISIS, however, the generation proces- ses can reacquire the necessary resources to pro- duce large evoked potentials because they have ‘recovered’ from their previous use. Although the underlying mechanisms for these effects are not known. similar results have been found with reac- tion time studies of successive stimulus presenta- tions and suggest that the time interval between task stimuli is an important determinant of processing outcomes (Kahneman, 1973; Keele, 1973; Kantowitz, 1974).

Given the influence of recovery cycle on sensory evoked potential amplitudes and the sensitivity of behavioral responses to the time intervals between stimulus presentations, it is reasonable to suppose that similar effects might be observed for the P3 potential as is implied by the presence of relatively small P3 components obtained with very short ISIS. If some sort of recovery cycle mechanism does contribute to P3 amplitude, changes in target stimulus probability and IS1 should interact with one another since the size of the P3 will vary inversely with the probability of the target stimu- lus but also will be affected by IS1 at some em- pirically defined limit. Just such an effect has been reported for visual stimuli: P3 amplitude de- creased as target probability increased (0.20 to 0.80) when IS1 was relatively short at 1.3 s, whereas

only a minimal change in P3 amplitude was ob- served across probability conditions when IS1 was relatively long at 3.0 s (Donchin, 1981; Donchin et al., 1986). However, another study which em- ployed auditory stimuli and manipulated target probability (0.10 to 0.30) with somewhat shorter ISIS (0.75 and 2.25 s) did not obtain this result (Fitzgerald and Picton, 1981). Although the two studies employed different stimulus modalities, since target probability and IS1 both appear to affect P3 amplitude, the exact nature of relation- ship between these variables is still unclear.

Because the target probabilities, stimulus pre- sentation rates, and other factors of these two reports were quite different, the present studies were conducted in order to assess the relationship between target probability and IS1 over a wide range of values while keeping the stimulus materi- als and task paradigm constant. If recovery cycle does contribute to component size, then P3 ampli- tude should vary inversely with target stimulus probability since frequently occurring targets will inhibit system recovery compared to infrequently presented targets. When IS1 is relatively long, however, target probability should have little ef- fect on P3 amplitude because the system resources will be recovered and can contribute fully to P3 generation regardless of how probable the stimu- lus may be. This hypothesis was evaluated in two different experiments in which target probability and IS1 were combined in separate blocks of trials so that the effects of just these variables could be assessed independently of possible sequence or timing factors factorially.

MATERIALS AND METHODS

Subjects Two different groups of 16 undergraduate stu-

dents (mean age 21.2, S.D. 1.8 years) from the University of California, San Diego were em- ployed in each experiment. All subjects reported no neurological or psychiatric problems, were naive to electrophysiological studies, and received course credit for their participation. Equal num- bers of each sex were used in each study.

165

Recording conditions Electroencephalographic (EEG) activity was re-

corded at the Fz, Cz, and Pz electrode sites of the lo-20 system using gold-plated electrodes affixed with electrode paste and tape. These were referred to linked earlobes with a forehead ground and impedance at 10 kS2 or less. Additional electrodes were placed at the outer canthus and supraorbit- ally to the left eye with a bipolar recording made of the EOG. The filter bandpass was OS-30 Hz (3 dB down, 12 dB octave/slope). The EEG was digitized at 3.0 ms per point for 512 ms with a 50 ms prestimulus baseline in Experiment 1, and 1.5 ms per point for 768 ms with a 75 ms prestimulus baseline in Experiment 2 (the different recording conditions for the IS1 manipulations were neces- sitated by equipment limitations). Wave forms were averaged on-line by a Nicolet Pathfinder II which also controlled the stimulus presentation and artifact rejection. All experimental conditions were recorded with eyes closed to minimize artifact (Polich, 1986a, 1987b), with trials on which the EEG or EOG exceeded a preset criterion rejected automatically. Subjects were highly cooperative in all task conditions which resulted in very few trials rejected due to artifactual responses.

Procedure ERPs were elicited with 1000 (standard) and

2000 Hz (target) tones presented binaurally at 60 dB SPL (9.9 ms, r/f, 50 ms, plateau) such that the target tone occurred randomly as appropriate for the specific probability/IS1 condition. Subjects were instructed to listen to the tones and raise the index finger of their right hand whenever a target was detected. The experimenter noted any misper- ceived trials. Each experiment employed the same general stimulus presentation conditions, with only the combination of probability and IS1 varied between studies.

A total of 20 artifact-free target stimulus trials were acquired for each combination of target probability and ISI. In each experiment the com- binations of different target probabilities and ISIS were such that the presentation order of the treat- ment conditions was counterbalanced across sub- jects and gender. Experiment I employed target probabilities of 0.10, 0.30, and 0.50 with ISIS of

1.5 and 3.0 s. Experiment 2 employed target prob- abilities of 0.20, 0.50, and 0.80 with ISIS of 4.0, and 10.0 s. Hence, each subject in each experiment received 6 different conditions presented in differ- ent blocks which were defined by the target stimulus probability and ISI.

RESULTS

All analyses of variance with repeated measures designs employed Greenhouse-Geisser corrections for inhomogeneity of covariance. Task perfor- mance for all experiments was virtually perfect, with a mean of 0.2 and 0.5 target tones misper- ceived across Experiments 1 and 2. respectively. A two-factor (probability x ISI) analysis of variance was applied to the performance data from each study. No significant effects of interactions were obtained for either experiment for the perfor- mance data.

Component measurement Wave forms from each electrode for each con-

dition and experiment were analyzed in the same fashion. The Nl-P2-N2-P3 complex was isolated first, the most negative and positive components corresponding to this morphology were assessed within the latency ranges of 80-120, 100-200, 160-240, and 240-400 ms, respectively. Ampli- tude was measured relative to the prestimulus baseline, with peak latency defined as the time point of maximum positive or negative amplitude. The amplitude and peak latency measures for the Nl, P2, and N2 components were obtained from the Cz electrode where they were largest. The grand averages for the ERPs obtained from each task condition ant electrode placement are pre- sented in Figs. 1 and 3 for Experiments 1 and 2. respectively (the shorter wave forms in Fig. 1 compred to Fig. 3 resulted from the differences in epoch length). The mean P3 amplitudes and latencies (fl S.E.) from the Pz electrode site plotted as a function of IS1 for each experiment are illustrated in Figs. 2 and 4. The generally small P3 components obtained in Experiment 1 ap- peared to originate from a sample of subjects with inherently low amplitude ERPs. As detailed be-

166

low, however, the typical effects of probability, ISI. and electrode site were obtained for the P3. Each set of P3 measures from each experiment was assessed with a 3-factor (probability x IS1 x

electrode) and separate two-factor (probability x

ISI) analyses of variance on the data from the Pz electrode. The amplitude and latency data from the Nl, P2, and N2 components for each experi- ment also were assessed with a two-factor (prob- ability X ISI) ANOVA.

P3 umplitude Experiment I. As is indicated by Figs. 1 and 2,

P3 amplitude was smaller for the shorter IS1 across probability conditions, even though decreases in

TARGET PROBABILITY

EOG w

Fig. 1. Grand averaged event-related potentials (n = 16) for the

0.10. 0.30. and 0.50 target stimulus probability conditions at the 1.5 and 3.0 s Inter-stimulus intervals for each electrode

position from Experiment 1.

10 30 50

TARGET PROBABILITY Fig. 2. Mean ( + 1 S.E.) P3 amplitudes and latencies from the Pz electrode site as a function of target stimulus probability for

each inter-stimulus interval condition of Experiment 1.

probability also produced declines in component amplitude. These observations were confirmed for all 3 electrode positions, with significant effects for target stimulus probability, F2.30 = 3.7, P -c

0.05, and IS1 obtained, F,,,5 = 9.4, P -c 0.005. The usually observed increase in P3 amplitude from the frontal to the parietal electrode sites also was found, F2,30 = 7.8, P -C 0.05, with electrode posi- tion and IS1 demonstrating a significant interac- tion, such that the 3.0 s IS1 produced somewhat stronger effects for electrode position compared to the 1.5 s ISI, F,,,,, = 5.3, P < 0.02. When each electrode site was analyzed separately, the 1.5 s ISI yielded significantly smaller P3 amplitudes thanthe3.0sISIat thePz(F,,,,= 14.7, PcO.002)

167

and Cz (F,,,, = 7.3, P -C 0.02) electrode sites, with a marginal effect obtained at the Fz recording position (F,,,, = 4.0, P = 0.07). Increases in target stimulus probability demonstrated a significant decrease in P3 amplitude only for the Pz electrode site, Fz,30 = 6.3, P < 0.01, although the interaction between probability and IS1 for P3 amplitude at the Fz electrode position was significant, F2,30 = 6.3, P < 0.01.

Experiment 2. Figs. 3 and 4 illustrate the effects of more extended probabilities and ISIS on P3 amplitude. In contrast to the results obtained with shorter ISIS in Experiment 1, the only significant effect observed in Experiment 2 was the usual increase in P3 amplitude from the frontal to parietal electrode positions, F2,)o = 40.6, P <

TARGET PROBABILITY

.20 .50 .80

+5 ‘T +T

LATENCY tmsec) Fig. 3. Grand averaged event-related potentials (n = 16) for the 0.20, 0.50, and 0.80 target stimulus probability conditions at

the 4.0 and 10.0 s inter-stimulus intervals for each electrode position from Experiment 2.

I __f_ 4 tee ISI - 10secISI I

il

2'0 5lo 8'0

350 1 0

ii

g 325- 4

5 '\ \ \

s '\ ----_ 5

5 300- 4

;\%

z

275 I I I 20 50 80

TARGET PROBABILITY Fig. 4. Mean (+l SE.) P3 amplitudes and latencies from the Pz electrode site as a function of target stimulus probability for

each inter-stimulus interval condition of Experiment 2.

0.001. All other main effects and interactions were non-significant (P > 0.25 in all cases). Application of the same two-factor ANOVA to data from each electrode substantiated this lack of statistical relia- bility for any of the independent variables.

P3 Latency Experiment I. The 3-factor ANOVA performed

on the peak latency data for the P3 component indicated that the 1.5 s IS1 conditions produced somewhat shorter P3 latencies than did the 3.0 s IS1 conditions, although this effect was not robust, F 1,15 = 4.4, P = 0.05. An overall decline in P3 latency from the frontal to parietal electrode sites also was found, F, 3o = 7.0, P < 0.01, as is typi-

16X

tally the case with the tapping task (Polich. 1986b, 1987b). No reliable effects for probability or any of the interactions were significant (P > 0.20 in all cases). Analysis of P3 latency at each electrode position separately yielded only a significant effect for ISI at the Pz recording site, F,,,, = 4.6, P < 0.05, with no other main effects or interactions obtained. Although this finding confirmed that observed for the overall analysis, as suggested by the lower portion of Fig. 2, the effects of the ISI on the target probability manipulations were not strong.

Experiment 2. P3 latency demonstrated a sig- nificant decrease as target probability increased, F z,l,j = 8.0, P < 0.01. as well as a marginally sig- nificant decrease from the frontal to parietal re- cording sites, FZ.7,j = 3.4, P = 0.05. No other relia- ble effects were obtained. Assessment of each elec- trode separately showed that the decline in latency with increases in target probability was significant for all 3 electrode sites (P -C 0.03, in all cases), with no effects for IS1 obtained.

NI, P2, and N2 components Experiment 1. A two-factor (probability X ISI)

ANOVA was applied to the amplitude and latency data of the Nl. P2, and N2 components from the target stimuli measured at the Cz electrode site. For the amplitude data, only Nl was affected by any of the independent variables, with the 1.5 set IS1 demonstrating significantly smaller amplitudes than the 3.0 s ISI, F,,,, = 8.4, P -C 0.02. No other reliable amplitude effects were observed for any of these components. For the lutencq’ data, Nl peak

latency, F,,,, = 4.5, P -C 0.05, and P2 peak latency, F 1.15 = 5.2, P -C 0.05, produced significantly shorter latencies for the 1.5 compared the 3.0 s IS1 conditions across probability levels. No other sig- nificant effects for any of the latency data for these components were obtained.

Experiment 2. The same analysis procedures were applied to the Nl, P2, and N2 components of the second experiment. For the umplitude data, the Nl component demonstrated significantly smaller amplitudes at the 4 s compared to 10 s IS1 condition, F,,,, = 20.3, P -C 0.001. P2 amplitudes were larger overall for the 10 s relative to 4 s IS1 conditions, F,,,5 = 5.5, P < 0.05, although this ef-

fect originated primarily from the 0.80 probability condition. No IS1 effects were found for N2 am- plitude, but increases in probability produced a decrease in amplitude, F,,,, = 4.9, P < 0.02. For the lutency data, P2 latency was slightly longer for the 10 s compared to 4 s IS1 condition, F,,,, = 5.5, P < 0.05, with no other significant latency results obtained.

DISCUSSION

P3 amplitude decreased with increases in target stimulus probability for the ISIS of 1.5 and 3.0 s but exhibited no difference for longer ISIS of 4.0 and 10.0 s. P3 latency generally was unaffected for the short ISIS of Experiment 1, but did decrease with increases in target probability when ISIS were longer and the probability range wider in Experi- ment 2. Given the ranges of probability, ISI, and the auditory modality employed, target stimulus probability affected P3 amplitude only when the time between stimulus presentations was compara- tively short, whereas peak latency was sensitive to target probability only when IS1 was long.

This relationship between target probability and IS1 suggests that the smaller P3 amplitudes ob- served previously with short ISIS (Woods et al.. 1980; Fitzgerald and Picton. 1981; Woods and Courchesne, 1986) may originate from limitations imposed on the P3 generating system in a manner similar to the ‘recovery cycle’ effects observed for sensory evoked potentials (Davis et al., 1966; Roth et al., 1976; Polich et al., 1988). If this is the case, then P3 amplitude is sensitive to variables other than those related to stimulus probability as has been suggested in other contexts (Johnson, 1986; Polich, 1987a). Although the present study does not rule out the possibility of ‘temporal probabil- ity’ as the underlying explanation for these find- ings, the observed target probability and IS1 ef- fects may have resulted from the increased de- mands placed on system resources from rapid stimulus presentation, rather than from the tem- poral frequency of target occurrence per se. In this view, P3 amplitude is reduced with high target probability at relatively short ISIS because the processing system is not fully recovered from the

169

recent component generation when the next target stimulus is presented. With low target probability and/or relatively long ISIS, the system is re- covered sufficiently to produce comparatively large P3 components since the target stimulus is pre- sented infrequently. These effects, in turn, may stem from limits on memory function such that the mechanisms responsible for the decay of the trace resulting from a previously processed target stimulus cannot become fully recovered with high probability stimuli or short ISIS (Squires et al., 1976; Donchin et al., 1986). Thus, the magnitude of P3 amplitude is affected by target probability and ISI because of limitations on processing capacity when stimuli must be evaluated in fre- quent succession.

Shorter P3 latencies generally were obtained when target stimulus probability was 0.80 com- pared to 0.20 in Experiment 2, a finding consistent with several previous ERP reports which have found that target stimuli presented with high probability are evaluated more quickly than low probability presentations (Campbell et al., 1979; Duncan-Johnson and Donchin, 1982; Brookhuis et al., 1983). This speeding of the evaluation pro- cess would appear to occur independently of those processes involved in P3 amplitude generation at least for relatively long ISIS. With shorter ISIS, P3 latency did not demonstrate consistent results, most likely because of the rapidity with which the peak latencies were produced in Experiment 1 relative to Experiment 2. In general, however, to the degree P3 latency reflects the speed with which target stimuli are processed, peak latency was affected only by the overall target stimulus prob- ability rather than by the temporal relationship to other stimuli as was P3 amplitude.

Although a processing resource interpretation of the observed probability and IS1 effects is speculative, this view also is suggested by several ERP studies which have found an interaction be- tween task difficulty and target stimulus probabil- ity. When target items are difficult to process, P3 amplitude is reduced and much less affected by overall probability than if target items are rela- tively easy to process even though task perfor- mance is equitable across conditions (Kramer et al., 1986; Polich, 1987a; Ruchkin et al., 1987).

This interaction between target probability and task difficulty indicates that while P3 amplitude reflects the fundamental operations of memory- updating processes (Donchin 1981; Klein et al., 1984; Donchin et al., 1986) it is also sensitive to the allocation of available processing resources used to perform the eliciting task (Isreal et al., 1980; Wickens et al., 1983; Kramer et al., 1985) Hence, when stimulus events occur frequently be- cause stimulus IS1 is short, more resources are consumed in a given amount of time with less frequently occurring events and relatively small P3 amplitudes are produced. When stimulus events occur infrequently (e.g. at ISIS of about 4.0 s or longer for auditory stimuli), the P3 generation system can recover more fully and relatively large P3 amplitudes are produced. By assuming that resource limitations can affect P3 amplitude, the effects of target stimulus probability and IS1 found in the present experiments may be related to the interaction between task difficulty and target probability that has been observed in other con- texts.

ACKNOWLEDGEMENTS

This work was supported by the Armstrong McDonald Foundation (formerly the J.M. Mc- Donald Foundation) and NIAAA Grant AA06420 and is publication number 6183NP from the Re- search Institute of Scripps Clinic.

REFERENCES

Brookhuis, K.A., Mulder. G., Mulder, L.J.M. and Gloerich,

A.B.M. (1983) The P3 complexes as an index of informa-

tion processing: the effects of response probability. Eiol. PsychoI., 17: 271-296.

Campbell, K.B.. Courchesne, E.. Picton, T.W. and Squires.

K.C. (1979) Evoked potential correlates of human informa-

tion processing. Bid. Psychol., 8: 45-68. Davis, H., Mast, T., Yoshie, N. and Zerlin, S. (1966) The slow

response of the human cortex to auditory stimuli: recovery

processes. Electroenceph. Clin. Neurophysiol., 21: 105-113. Donchin, E. (1981) Surprise!. Surprise? (1981) Psychoph_vsl-

ologv, 18: 493-513. Don&in, E. and Coles, M.G.H. (1988) Is the P300 component

a manifestation of context updating? Behav. Brarn Sci., 11: 357-374.

170

Donchm, E., Karis, D.. Bashore. 7. and Coles. M.G.H. (1986) Cognitive psychophysiology and human information

processing. In M.G.H. Coles. E. Donchin and SW. Porges.

( Eds.). Psychoph),siologv: Systems Processes and Applica-

r/ens. The Guilford Press, New York. pp. 244267.

Duncan-Johnson, C.C. and Donchin, E. (1977) On quantifying

surprise: the variation in event-related potentials with sub-

Jective probability. Psychophysrologv, 14: 456-467.

Duncan-Johnson. C.C. and Donchin, E. (1982) The P300 com- ponent of the event-related brain potential as an index of

information processing. Biol.Ps_ychol.. 14: l-52.

Fabiani. M., Karis, D. and Donchin. E. (1986) P300 and recall

in an incidental memory paradigm. Psychophyslolo~. 23:

298-314.

Fitzgerald, P.G. and Picton, T.W. (1981) Temporal and

sequential probability in evoked potential studies. Can. J.

P.rychol.. 35: 18X-200.

Fitzgerald. P.G. and Picton, T.W. (1984) The effects of prob- abihty and discriminability in the evoked potential to un-

predictable stimuli. In R. Karrer, J. Cohen and P. Tueting

(Eds.). Bruin and Information: Event-Related Potentrals.

Ann. N.Y. Acad. Ser., 425: 199-203.

Isreal. J.B., Wickens, CD.. Chesney, G.L. and Donchin. E.

(1980) The event-related brain potential as an index of

display-monitoring workload. Hum. Factors, 22: 211-224.

Johnson, R. (1986) A triarchic mode1 of P300 amplitude.

Ps,,chophysiologv. 24: 367-384.

Kahneman. D. (1973) Attention and Effort. Prentice-Hall. Inc.,

Englewood Cliffs, NJ, pp. 162-177.

Kantowitz, B.H. (1974) Double stimulation. In B.H. Kanto-

witr. Human Informatwn Processing: Tutorruls rn Perfor-

nwnce and Cognrtron. Lawrence Earlbaum Associates. Hil-

Isdale. NJ, pp. X3-131.

Karis. D., Fabiani. M. and Donchin. E. (1984) ‘P300’ and memory: individual differences in the von Restorff effect.

Cogn. PsychoI.. 16: 177-216.

Keele, S.W. (1973) Attention and Human Performmce.

Goodyear Publishing Company, Inc., Pacific Palisades, CA

pp. 977107.

Klein. M.. Coles. M.G.H. and Donchin. E. (1984) People with absolute pitch process tones without producing a P300.

Science, 223: 1306-l 309.

Kramer. A.F.. Wickens, C.D. and Donchin, E. (1985) Processing properties: evidence for dual-task integrality. J.

Exp. PsychoI.: Hum. Percept. Perfor., 11: 393-40X.

Kramer. A.F., Schneider, W., Fisk and Donchin, E. (1986) The effects of practice and task structure on components of the

event-related brain potential. P.sychophysiolog)/. 23: 33-42.

Neville, H.J., Kutas, M., Chesney. G. and Schmidt, A.L. (1986) Event-related brain potentials during initial encoding and recognition memory of congruous and incongruous words.

J. Mem. I,ang., 25: 75592.

Paller. K.A., McCarthy. G. and Wood, C.C. (1988) ERPs predictive of subsequent recall and recognition perfor-

mance. BIoI. Psvchol.. 26: 269-276.

Picton, T.W. and Stuss. D.T. (1980) The component structure of the human event-related potential. In H.H. Kornhuber

and L. Deeke (Eds.). MotiLution, Motor, and Sensor,

Processes of the Bruin, Progress in Bruin Research. Vol. 54,.

Elsevier/North Holland Biomedical Press, Amsterdam, pp.

17-49.

Polich, .I. (1986a). Attention, probability. and task demands as

determinants of P300 latency from auditory stimuli. Elec-

troenceph. Clin. Neurophysiol.. 63: 251-259.

Polich, _I. (1986b) Normal variation of P300 from auditory

stimuli. Electroenceph. Clin. Neuroph.vslol.. 65: 236-240.

Polich, J. (1987a) Task difficulty, probability, and mter-stimu-

lus interval as determinants of P300 from auditory stimuli.

Electroenceph. Clin. Neurophysrol.. 6X: 31 l-320.

Polich, J. (1987b) Response mode and P300 from auditory

stimuli. Biol. Ps_vchol.. 25: 61-71.

Polich, J.. Howard. L. and Starr. A. (1983) P300 latency

correlates with digit span. Psychophysrolqv. 20: 665-669.

Polich, J.. Aung. M. and Dalessio. D.J. (1988) Long latency

auditory evoked potentials: intensity, inter-stimulus inter- val, and habituation. Puolouian J. Biol. Sci.. 23: 35-40.

Roth, W.T., Krainz. P.L., Ford, J.M., Tinklenberg, J.R..

Rothbart, R.M. and Kopell. B.S. (1976) Parameters of

temporal recovery of the human auditory evoked potential.

Electroenceph. Clan. Neuroph_wol.. 40: 623-632.

Ruchkin, D.S., Sutton, S. and Mahaffey, D. (1987) Functional

differences between members of the P300 complex: P3e

and P3b. PsvchophyGolow. 24: 87-103.

Squires, K.C., Wickens, C.. Squires, N.C. and Donchin. E.

(1976) The effect of stimulus sequence on the waveform of

the cortical event-related potential. Science, 193: 1142-

1146.

Sutton, S. (1979) P300-thirteen years later. In D. Lehman

and E. Callaway (Eds.), Evoked Bruin Potentials and Beha-

Ior, Plenum Press. NY, pp. 107-126.

Sutton, S.. Braren. M., &bin. J. and John, E.R. (1965) Evoked

potential correlates of stimulus uncertainty. Scrence, 150:

118771188.

Tueting, P.. Sutton, S. and Zubin. J. (1971) Quantitative evoked

potential correlates of the probability of events. P.yvcho-

ph.vsiologv, 7: 3X5-394.

Wickens, C., Kramer. A., Vanasse, L. and Donchin. E. (1983)

The performance of concurrent tasks: a psychophysiologi- cal analysis of the reciprocity of information processing

resources. Science, 221: 1080@1082.

Woods, D.L. and Courchesne. E. (1986) The recovery function

of auditory event-related potentials during split-second dis- criminations. Electroenceph. C/in. Nruroph~vsiol., 65: 304-

315.

Woods, D.L., Courchesne, E., Hillyard, S.A. and Galamhos, R. (1980) Recovery cycles of event-related potentials in multi-

ple detection tasks. Electroenceph. Clm. Neuroph.v.riol. 1 50:

335-347.