ERPs to stimuli requiring response production and inhibition: effects of age, probability and visual noise

Electroencephalography and clinical Neurophysiology, 1988, 71:55-63

55

Elsevier Scientific Publishers Ireland, Ltd. EEG 03344

ERPs to stimuli requiring response production and inhibition: effects of age, probability and visual noise Adolf Pfefferbaum and Judith M. Ford Veterans Administration Medical Center, Palo Alto, CA 94304 (U.S.A.), and Department of Psychiatry and Behavioral Sciences, Stanford University Medical Center, Stanford, CA 94305 (U.S.A.) (Accepted for publication: 17 March 1987)

Summary Sixty-six normal adults ranging in age from 20 to 85 years were presented with stimuli containing explicit instructions to initiate or to inhibit a motor response (the words 'push' or 'wait'). In one task, the effect of stimulus probability was investigated by varying probability between 0.25 and 0.75 for both Go and No-go stimuli. In another task, the effect of visual noise was investigated by degrading the stimuli with ampersands on half of the trials. Regression analysis was used to examine the effects of age on P3 amplitude and latency for each stimulus type. The effects of stimulus variables on P3, independent of age, were examined by standardizing each subject's data to tlfose expected for a 20 year old. P3 latency to all stimuli and RT to Go stimuli increased with age. The latency of P3s to No-go stimuli was less sensitive to age than Go stimuli. P3 amplitude at Cz and Pz (but not Fz) diminished with age. P3s to Go stimuli were maximal at Pz and earlier than P3s to No-go stimuli. P3s to No-go stimuli were maximal at Cz. These differences between Go and No-go stimuli remained true under visual noise and probability manipulations. Visual noise prolonged the latency of Go and No-go P3. Less probable Go and No-go stimuli elicited larger and later P3s than more probable stimuli. Decreasing the probability of the No-go stimulus enhanced its central distribution. Key words: Event-related potential; Go stimuli; No-go stimuli; P3 latency

We have recently reported the results of a study using stimuli containing explicit instructions to initiate or to inhibit a motor response (the words 'push' or 'wait') to elicit ERPs (Pfefferbaum et al. 1985). These stimuli might have several advantages for use with clinical populations: (1) they explicitly contain the task instructions and thus require no translation by subjects; (2) they both evoke large endogenous components efficiently; and (3) they elicit P3s with distinct topographies and latencies. However, before use in clinical populations, age norms are needed, as well as

1 This research was supported by the Medical Research Service of the Veterans Administration and NIAAA Grant AA05965.

Correspondence to: Dr. Judith M. Ford, Dept. of Psychiatry and Behavioral Sciences, Stanford University Medical Center, Stanford, CA 94305 (U.S.A.).

further explorations of which stimulus parameters are optimal. This paper presents a study in which 'push' and 'wait' stimuli, presented under conditions of varying probability and visual noise; were used to test healthy subjects across the adult age range. 'Push' and 'walt' are explicit and direct examples of Go and No-go stimuli. No-go P3s are typically later, smaller, and more anteriorly distributed across the midline scalp than Go P3s (Hillyard et al. 1976; Tueting and Sutton 1976; Simson et al. 1977; Pfefferbaum et al. 1984). The topographic difference is the principal Go/No-go effect. In our research, this Go/No-go effect was found regardless of whether the Go and No-go stimuli were symbolic or linguistic, whether the stimuli were degraded with additional characters, whether subjects made a button press or simply counted the Go stimulus, or whether it was ob-

0168-5597/88/$03.50 © 1988 Elsevier Scientific Publishers Ireland, Ltd.

56 tained on the first or the second presentations of the sequences (Pfefferbaum et al. 1985). The G o / N o - g o topographic effect is similar to that seen with changes in probability in the graphic data inFig, lb published by Duncan-Johnson and Donchin (1977). In this study, frequent targets elicited P3 with a flat distribution across Cz and Pz while infrequent targets elicit P3s with a more parietal distribution. The distribution of P3 to frequent stimuli is thus similar to the distribution of P3 to 'wait' stimuli. The apparent topographic effects of probability, as well as a desire to maximize the amplitude of the parietal No-go P3 motivated the use of a probability manipulation in the current study. P3s to No-go stimuli are also later than P3s to Go stimuli (Simson et al. 1977; Pfefferbaum et al. 1980, 1984, 1985). Older subjects, who have slower P3s than younger subjects in a wide range of cognitive tasks (Ford and Pfefferbaum 1986), also show additional P3 latency increase to No-go stimuli (Pfefferbaum et al. 1980). In a study using the visual modality, and testing subjects over the entire adult age range (Pfefferbaum et al. 1984), we found that P3 latency was delayed 1.15 msec/year to the Go stimulus, but 1.48 msec/year to the No-go stimulus (i.e., non-target rare). We have speculated that this may be a subtle manifestation of an age-related deficit in response withholding. The current study was intended to determine if the P3 slowing for No-go stimuli in the elderly will still obtain when words ('push' and 'wait'), rather than symbols, are used as stimuli. Finally, delay in P3 latency with age could be due to more time being taken for the cognitive operations responsible for the generation of the particular P3. Salthouse (1978) has shown that age-related slowing in reaction time is exaggerated with more difficult cognitive tasks. We employed visual degrading, a direct manipulation of task difficulty, to evaluate the contribution of this variable to increase P3 latency with age. Methods

Subjects Sixty-six community volunteers spanning the adult age range were tested. Subjects were re-

A. PFEFFERBAUM,J.M. FORD cruited and screened, by phone interview and questionnaire, to exclude those with a history of significant psychiatric or neurological histories, recent use of psychoactive drugs, or alcohol consumption exceeding 50 g/day. The age distribution of the subjects was 20-25 years = 8; 26-35 years = 12; 36-45 years = 12; 46-55 years = 6; 56-65 years = 10; 66-75 years = 16; and 76-85 years = 2.

Stimuli The words 'push' and 'wait' were presented on a display CRT for 1200 msec at 2 sec intervals in random order. Subjects were instructed to watch the stimuli and follow the instructions, paying equal attention to speed and accuracy of response. In run 1, the effect of visual noise was investigated by presenting the words 'push' and 'wait' with equal probability, but for half of the trials, the words were degraded with ampersands, e.g., & P & U & S & H & and & W & A & I & T & . The 4 stimuli, each occurring with a probability of 0.25, will be referred to as PUSH, &PUSH, WAIT, and &WAIT. The order of stimulus presentation was random. In runs 2 and 3, stimulus probability was varied. In run 2, PUSH appeared 25% of the time (25 PUSH) and WAIT 75% of the time (75 WAIT). In run 3, WAIT appeared 25% of the time (25 WAIT) and PUSH 75% of the time (75 PUSH).

Recording EEG was recorded from Fz, Cz, and Pz referenced to linked earlobes; EOG was recorded from electrodes above and below the right eye. The amplifier bandpass was 0.03-100 Hz. EEG and EOG were sampled every 5 msec for a 1230 msec record, with a prestimulus baseline of 50 msec. Reaction times (RTs) were recorded for each stimulus presentation. Data for each trial were saved for off-line analysis.

Data analysis EEG trials contaminated with EOG (EOG greater than + / - 50/~V) or E M G (EMG greater than + / 100 gV) artifact were rejected automatically before further processing. RTs longer than 1230 msec were considered errors of omission (Misses). Responses to the No-go stimuli

57

ERPs TO RESPONSE PRODUCTION AND INHIBITION STIMULI were considered errors of commission (False Alarms). Stimulus synchronized ERPs to the various stimuli were constructed for qualitative evaluation and are presented in Figs. 1 and 2. However, for data quantification, latency-adjusted ERPs were constructed. The procedure for constructing these latency-adjusted averages, originally proposed by Woody (1967), has been used before (Pfefferbaum et al. 1984). Recent improvements to enhance the 'goodness of fit' and guard against bias introduced by low signal to noise ratios are described below. The digitized EEG recorded from each lead is reduced to a 10 msec sample rate by omitting every other point and applying a low pass digital filter at 3.5 Hz (down 3 dB) (Ruchkin and Glaser 1978). A 2.0 Hz half sine wave is used as the initial template. It is moved across each trial in 10 msec increments over 2 latency ranges: 300-690 msec ('signal range') and 700-1090 msec ('noise range'). The covariance between the template and the corresponding points of the single trial are calculated at each positioning, and the latency in the signal range at which the largest covariance is found is tentatively identified as P3 latency for that trial. The covariance calculation is used instead of correlation because the former gives weight to both wave amplitude and shape whereas the latter considers only wave shape in determining the best fit. The tentative P3 signal is only accepted as a bona fide signal if it meets 2 criteria: its covariance must be larger than the maximum covariance in the noise range on that trial, and its wave shape must correlate with the template with r > 0.36 (P < 0.05, one-tailed test, for the number of points in the correlation). If both conditions are not met, the trial is deemed to lack an identifiable P3 and is rejected from further analysis. The remaining single-trial P3 latency estimates are used to calculate the median P3 latency value for infrequent stimuli, correlations between P3 latency and RT on individual trials, and to construct latency-adjusted averages. A latency-adjusted average is one in which each individual trial has been latency-adjusted before summation. The adjustment consists of shifting each trial so that the latencies estimated for P3 (using the template matching procedure described above) on each trial

co-occur at 300 msec (an arbitrarily selected latency, consistent with the accepted nomenclature 'P300'). P3 amplitude is measured as deviation from pre-stimulus baseline at this point of alignment. The resulting amplitude and latency data were analyzed for age effects by means of regression analyses with age as the independent variable (SPSS 1975). The effects of the various manipulations on P3 (Go/No-go, visual noise, and probability) independent of age were examined by standardizing each subject's data to those expected for a 20 year old using the appropriate age regression values. The G o / N o - g o effect on the midline distribution of P3 amplitude was measured as the difference between P3 amplitude at Pz and Cz, referred to as P3 (Pz-Cz). The single-trial correlation of P3 latency with each trial's RT was obtained for each subject for each condition and z-transformed. The P 3 / R T correlation was regressed against age for each stimulus.

Statistical analysis The effects of age on all measures were estimated by linear regression analyses and expressed as slopes (#V/year or msec/year) and correlation coefficients. To determine if the effects of age on P3 were greater to one stimulus than to another (i.e., comparison of slopes), the difference between the responses to two stimuli (e.g., &PUSH-PUSH) was regressed against age. The effects of experimental variables were estimated using pairwise t tests, both before and after the effects of age were removed using the procedure described above.

Results

The data were analyzed, first for the effects of age, and then for the effects of the experimental variables with the effects of age removed. Each analysis is presented separately below. The grand averages by decade are presented in Figs. 1 and 2.

Age P3 latency.

P3 latency was regressed against

58

A. P F E F F E R B A U M , J.M. F O R D

&P&U&S~H~

PUSH

&kl&A&

NAIT

i

AGE

¢T&

L

20-29

30-39~ ~

30-39

qO-q9

5O-59

q0-q9

[ ~

~

r

~ " _ ~

.

i

50-59

i

i~

I! ~

i

PZ CZ . . . . . . . . . . .

-,

PZ CZ ............

60-69

-

-

~

...............

_

.

~

6D-59

0uV

70-81

70-8]

[

i o

~oo

~oo

qou

.... '

~ou

MS~E I

Fig. 1. Stimulus-synchronized grand average ERPs by decade for regular ( P U S H and W A I T ) and degraded (&P&U&S&H& and &W&A&I&T&) stimuli.

age for each stimulus (Figs. 3 and 4). The associated correlations and slopes are presented in Table I. The largest age correlation and steepest slope were obtained for &PUSH (r = 0.61, slope = 1.79 msec/year). Comparison of slopes indi-

cates that No-go stimuli elicit P3s less sensitive to age than do Go stimuli (i.e., smaller correlations and less steep slopes), but the difference is only significant for P3s elicited by &PUSH and &WAIT (P < 0.05). The P 3 / R T correlation was also re-

TABLE I Age regression analysis and slope comparisons (msec/year) for P3 latency, reaction time (RT), and single-trial P 3 / R T corr. Stimulus

PUSH &PUSH WAIT &WAIT 75 P U S H 75 W A I T 25 P U S H 25 W A I T * P < 0.05.

P3 latency (Pz)

RT

P 3 / R T corr.

Corr.

Slope

Corr.

Slope

0.55 0.61 0.36 0.24 0.51 0.36 0.48 0.44

1.42 1.79 0.92 0.67 1.24 0.98 1.06 1.03

0.44 * 0.40 *

1.98 2.34

-0.25 * -0.21 *

0.43 *

1.68

-0.23 *

0.28 *

1.54

-0.20

* * * * * * * *

ERPs TO RESPONSE P R O D U C T I O N A N D INHIBITION STIMULI PUSH 2 5 %

NAIT

59 NAIT

PUSH 757.

757.

257.

AGE

flGE

i

20-29

20-29 I I I

30-39

30-39

~

qO-q9

~

q0-q9

I

E

i

~ " ~ : ' : : ' ~

50-59

r" ~

I

50 - S 9

! Pz

/'~%..'~

"

! .

.

.

.

.

.

.

.

.

.

.

.

PZ

j

CZ. . . . . . . . . . . .

I : ~ 60-69

60-69

lo~v

70-81

70-81

i

i 0

i yu0

uuu

M~t

Fig. 2. Stimulus-synchronized grand average ERPs by decade for low (25%) and high (75%) probability stimuli. &P&U&S&H&

PUSH 600

6O0

M S E C

• =

•

500

"=i . . . .

'=1"

."

" k" " ""

"

I"

•

"I

"

. . .'- -

400

400

300

300

I

I

I

I

20

30

40

50

60

I

I

70

80

I

"

"

'

"

~''

•

I

••~"

t

I

I

I

I

I

20

30

40

50

60

70

WAIT

"

'

"

"

80

&W&A&I&T&

600

M S E C

"

500

"

60O .....

it

•

" "

il•"

•1

"

.........

....

...

,'.:.

500 bl 41]0

P

400

300

•

•

II ~

•

••

•

5"7" 7":..""

.....

I

I

I

I

I

I

I

20

30

40

50

60

70

80

l,

,,

......

300

I

I

I

I

I

I

I

20

30

40

50

60

70

80

I

I

AGE AGE Fig. 3. P3 latency (median, iteration 1, single-trial analysis) plotted as a function of age. The solid lines are the predicted regression function. The dotted lines are 1 and 2 S.D. from the predicted. Data are for the regular and degraded stimuli.

60

A.

P U S H 75% 600

_

S E

500

_

C

400

PFEFFERBAUM,

J.M.

FORD

P U S H 25% 600

M • . . Y..

" ' b ", .....

500

. . . . . . . .

=.=_ .

.-;-'" _

.

.....

; ....

.

.

.

:..; - . .

,00

• ,'r . .-

300

I

i: ~- . . . . . . . .

. . .

=

y

4

300

I

I

t

I

t

I

I

20

30

40

50

60

70

80

--

I

I

I 20

30

W A I T 75% 600

..

I

f

I

f

I

40

50

60

70

80

I

W A I T 25%

--

600~_

M S

500

C

............

• •

E

"

"

u-, ~ . . .

•

k

;i "="

,®

"

. --

400

400

]]i'[']]![i']]'[~ii/

• ....

•

. . . . . . . . . . . .

•

•

.....

300

.

. _ . . . . . . . i r

•

-

~_',~

•

....

, -. ."

=_

. . . . . . .

,"

....

...... •

o"

"

•

~ ....

"

=

•

". . . . . . . . . .

300

I

1

I

I

I

I

20

30

40

50

60

I 70

I

i

80

I

I

I

I

I

I

I

20

30

40

50

60

70

80

AGE

AGE

Fig. 4. S a m e as Fig. 3 f o r l o w a n d h i g h p r o b a b i l i t y stimuli.

gressed against age for each G o stimulus. There was a tendency for the single-trial P 3 / R T correlation to decrease with age (Table I). P3 amplitude. Correlations between age and P3 amplitude were computed separately for each electrode. At Cz and Pz the amplitude of P3 diminished with increasing age, but at Fz it did not. The difference between P3 amplitude at Pz and Cz (P3 (Pz-Cz)) served as a measure of the G o / N o - g o scalp topography effect and was not affected by age for any of the stimuli. Performance measures. Age was regressed against RT to each Go stimulus. Table I statistics indicate that while RT to &PUSH increased 2.34 msec/year, RT to PUSH increased only 1.98 msec/year, this difference was not significant, however. Subjects made almost no errors on these tasks. When the numbers of errors were regressed against age none of the correlations was significant, although they were all either zero or positive, suggesting that subjects make more errors with increasing age.

Experimental variables The following results are based on analysis of values which have been standardized to age 20. The effects of the experimental variables were also examined before the data were standardized to age 20. There were no substantive differences between the two methods of analysis, suggesting that there were no significant interactions of the experimental variables with age. Go vs. No-go. Go stimuli evoked a P3 with a Pz maximum and No-go stimuli evoked a P3 with

T A B L E II T h e G o / N o - g o t o p o g r a p h y effect r e f l e c t e d in t h e P3 (Pz-Cz) d i f f e r e n c e score. Comparison

t

P U S H vs. W A I T & P U S H vs. & W A I T 75 P U S H vs. 75 W A I T 25 P U S H vs. 25 W A I T

7.92 8.62 8.43 10.33

• * * P < 0.001, t w o - t a i l e d test.

*** *** *** ***

ERPs TO RESPONSE PRODUCTION

P3

AMPLITUDE

PUSH 20 15

&P&U&S&H&

WAIT

~----~

d /

10

10

5

5 I

I

Fz

&W&A&I&T&

15

r/ /

Cz

I

I

I

I

Pz

Fz

Cz

Pz

PUSH25% /

20

-,

WArT25~

20

15

15

10

10

5

5

61

AND INHIBITION STIMULI

PUSH75c;# WAIT75%

I

I

I

I

I

I

Fz

Cz

Pz

Fz

Cz

Pz

Fig. 5. M e a n a g e - a d j u s t e d P3 a m p l i t u d e s f r o m the l a t e n c y - a d j u s t e d E R P s for the r e g u l a r a n d d e g r a d e d , a n d the low a n d • high p r o b a b i l i t y stimuli.

a Cz maximum. This topographic effect is illustrated in Fig. 5 and can be seen to hold up under visual noise and probability manipulations.

All G o stimuli evoked a larger P3 (Pz-Cz) than did No-go stimuli. Table II lists the results of the t tests. P3 to No-go stimuli was later than to G o stimuli. This effect held up under both the visual noise manipulation and the probability manipulation. P3 to No-go stimuli was later than to G o stimuli for both regular, by 50 msec (t = 10.66, P < 0.001) and degraded, by 70 msec (t = 11.55, P <0.001) conditions. Response inhibition delayed P3 in both probable (75 W A I T vs. 75 PUSH) and improbable (25 W A I T vs. 25 PUSH) comparisons. Table III lists latency values for all the P3s and t values for the various pairwise comparisons. Visual noise effect. Degrading the stimulus reduced the amplitude of P3 to G o but not to No-go stimuli, as seen in Table III. Degrading the stimulus delayed P3 for both G o (t = 4.34, P < 0.001) and No-go (t = 6.92, P < 0.001) stimuli. Degrading the stimulus delayed RT by about 40 msec (t = 5.27, P < 0.001), also seen in Table III. Probability effect. Less probable G o and Nogo stimuli (25 P U S H and 25 WAIT) elicited larger and later P3s than their more probable analogs (75 P U S H and 75 WAIT). This was seen at all 3 leads. Decreasing the probability of the No-go stimulus enhanced the distinctive scalp distribution of P3 to No-go stimuli; t h e Pz-Cz difference was more negative (that is, Cz > Pz) for 25 W A I T than to 75 W A I T (t = 4.34, P < 0.001). RT to 75 P U S H was 44 msec faster than the R T to 25 P U S H (t = 5.79, P < 0.001) (Table III).

TABLE III A g e - c o r r e c t e d m e a n a m p l i t u d e s a n d latencies o f P3, r e c o r d e d at Pz, a n d r e a c t i o n t i m e for each s t i m u l u s type. P3 a m p l i t u d e (/~V)

P3 l a t e n c y ( m s e c )

R e a c t i o n time ( m s e c )

PUSH

WAIT

PUSH

WAlT

PUSH

22.7 21.2 t = 3.29 *

18.4 18.2 t = 0.43

t = 8.12 * t = 5.22 *

378 397 t = 4.34 *

428 t = 10.66 * 468 t = 11.55 * t = 6.92 *

399 437 t = 5.27 *

24.2 17.7 t =12.15 *

21.4 t = 3.76 * 15.3 t = 4.83 * t =11.08 *

396 381 t = 3.75 *

435 t = 8.69 * 399 t = 3.63 * t = 7.20 *

410 366 t = 5.79 *

Visual noise Regular Degraded

Probability 25% 75%

* P < 0.002, t w o - t a i l e d test.

62 Discussion

P3 to a stimulus instructing a subject not to make a response (No-go) is different in several ways from P3 to a stimulus instructing the subject to make a response (Go). The No-go P3 is smaller, later and has equal amplitudes at central and parietal electrode sites. This contrasts to the parietally maximal distribution of P3 to Go stimuli. In a previous report we found that this difference in topography was robust with respect to the type of response required (counting vs. button pressing), the addition of visual noise, and the type of stimulus presented (words vs. symbols) (Pfefferbaum et al. 1985). Neither the Pz-Cz scalp distribution of P3 to Go nor to No-go stimuli were affected by age in the current study, providing evidence that this topography effect is also robust to the effects of age. This is not inconsistent with the general age-related topography effects reported by us and others (e.g., Pfefferbaum et al. 1980; Tecce et al. 1980) where the age-effect is large at Fz and is greatly influenced by the slow wave. One of the purposes of this study was to determine whet'her the scalp distribution of P3s to the Go and the No-go stimuli changed as the stimuli became more or less surprising. The data illustrated by Duncan-Johnson and Donchin (1977) indicated that as target stimuli became more frequent, they elicited P3s with a more central distribution. In the current study, the less frequent No-go stimuli (25 WAIT), elicited a more centrally distributed P3 than the more frequent No-go stimuli (75 WAIT). Probability had no effect on the Pz-Cz measure of distribution of the G o P3. This study was also designed to investigate how P3 latency and RT would be affected by adding visual noise to the stimuli, and how this change would be manifest across the age range. Degrading the stimuli in this way delayed RT an average of 55 msec: a magnitude of change comparable to an earlier study (Pfefferbaum et al. 1985). Of particular interest was how this effect interacted with age. Age-regressions against RT for PUSH and &PUSH were compared. RT to PUSH increased 1.98 m s e c / y e a r while RT to

A. PFEFFERBAUM, J.M. FORD &PUSH increased 2.34 msec/year. However, while this difference in slopes was in the predicted direction, it was not significant. The effect of degrading on P3 latency was similarly affected by age. While P3 latency to &PUSH increased 1.79 m s e c / y e a r and only 1.42 m s e c / y e a r to PUSH, this difference in slopes was also not significant. The same trend was not seen for the WAIT stimulus for P3 latency. While these data do suggest that in some situations age-related slowing in RT and P3 latency is exaggerated with more difficult tasks, it is weak because the trend was neither significant nor was it seen to the WAIT stimulus. In two previous studies, we observed that the P3 delay with age was greater to No-go than to Go stimuli (Pfefferbaum et al. 1980, 1984). We suggested that this effect may have been due to an age-related deficit in the response withholding process. However, in the data reported here P3 latency to the No-go stimuli was less affected by age than was P3 to the Go stimuli. Alternatively, the earlier finding might have been due to the elderly having difficulty decoding a symbolic Nogo stimulus. However, in the earlier experiment in young subjects, there was no evidence for symboric stimuli being more difficult to decode than linguistic stimuli (Pfefferbaum et al. 1985). The issue may be clarified by comparison of P3 latency to linguistic and symbolic, Go and No-go stimuli in young and old subjects. The extent to which non-target stimuli actually involve response inhibition depends greatly upon the subject's interpretation of the task assigned, and the facility with which they translate the No-go instruction implicit in the non-target stimuli. Thus, the use of explicit verbal instructions to 'push' or to 'wait' allows a direct comparison of brain potentials associated with response production and inhibition. While it requires some cooperation, the task is easy to understand and can be performed at a high level of accuracy, which can be monitored. In this experiment, large endogenous components were elicited by these stimuli, which showed distinctive topographies. Decreasing the probability of No-go stimuli enhanced P3 amplitude and particularly emphasized its distinctively central distribution. We tentatively suggest that P3 to Go stimuli is

ERPs TO RESPONSE PRODUCTION AND INHIBITION STIMULI d i f f e r e n t f r o m P3 to N o - g o s t i m u l i a n d i n v o l v e s different generators. A larger recording montage might afford better estimation of the differences b e t w e e n t h e s e P3s. The authors wish to thank Ann Doty for testing subjects, Patsy White for analyzing data, and Margaret Rosenbloom for helping to prepare this report.

References Duncan-Johnson, C.C. and Donchin, E. On quantifying surprise: the variation of event-related potentials with subjective probability. Psychophysiology, 1977, 14: 456-467. Ford, J.M. and Pfefferbaum, A. Age-related changes in eventrelated potentials. In: P.K. Ackles, J.R. Jennings and M.G.H. Coles (Eds.), Advances in Psychophysiology. JAI Press, Greenwich, CT, 1986: 299-339. Hillyard, S.A., Courchesne, E., Krausz, H.I. and Picton, T.W. Scalp topography of the P3 wave in different auditory decision tasks. In: W.C. McCallum and J.R. Knott (Eds.), The Responsive Brain. John Wright, Bristol, 19-76: 81-87. Pfefferbaum, A., Ford, J.M., Roth, W.T. and Kopell, B.S. Age-related changes in auditory event-related potentials. Electroenceph. clin. Neurophysiol., 1980, 49: 266-276. Pfefferbaum, A., Roth, W.T., Ford, J.F., Wenegrat, B. and Kopell, B.S. Clinical application of the P3 component of event-related potentials. I. Normal aging. Electroenceph. clin. Neurophysiol., 1984, 59: 85-103.

63

Pfefferbaum, A., Ford, J.M., Weller, B.J. and Kopell, B.S. ERPs to response production and inhibition. Electroenceph. din. Neurophysiol., 1985, 60: 423-434. Ruchkin, D. and Glaser, E. Simple digital filters for examining CNV and P300 on a single-trial basis. In: D.A. Otto (Ed.), Multidisciplinary Perspectives in Event-Related Brain Potential Research. U.S. Environmental Protection Agency, Washington, DC, 1978: 579-581. Salthouse, T.A. The role of memory in the age decline in digit-symbol substitution performance. J. Gerontol., 1978, 33: 232-238. Simson, R., Vaughan, H.G. and Ritter, W. The scalp topography of potentials in auditory and visual Go/No-go tasks. Electroenceph. clin. Neurophysiol., 1977, 43: 864-875. SPSS, Statistical Package for the Social Sciences, McGraw Hill, New York, 1975:320 pp. Tecce, J.J., Yrchik, D.A., Meinbresse, D., Dessonville, C.L. and Cole, J.O. CNV rebound and aging. In: H.H. Kornhuber and L. Deecke (Eds.), Motivation, Motor and Sensory Processes of the Brain: Electrical Potentials, Behaviour and Clinical Use. Progr. Brain Res., Vol. 54. Elsevier, Amsterdam, 1980: 552-573. Tueting, P. and Sutton, S. Auditory evoked potential and lift/no-lift reaction time in relation to uncertainty. In: W.C. McCallum and J.R. Knott (Eds.), The Responsive Brain. John Wright, Bristol, 1976: 71-75. Woody, C.D. Characterization of an adaptive filter for the analysis of variable latency neuroelectric signals. Med. biol. Engng Comput., 1967, 5: 539-553.