Effects of warning-signal duration on the early and late components of the contingent negative variation

Biological Psychology, 3, 1975,X3-275.

EFFECTS

OF WARMNG-SIGNAL

AND LATE COMPONENTS NEGATIVE

RAFAEL

@ North-HoliandAtblishing

DURATION

Company

ON THE EARLY

OF THE CONTINGENT

VARIATION

KLORMAN

and ELLEN BENTSEN

Department of Psychology,

University of Rochester, Rochester, New York, U.S.A.

Accepted for p~lblication 15 August 1975

Slow EEG potentials were recorded from three sites on the scalp (Fz, Cz and Pz) during a simple reaction time task in which the duration of the warning signal was either 0.5 or 2 sec. The duration of the foreperiod was held constant, and order of conditions was varied according to a Iatin square design. As predicted, the longer warning signat evoked increased amplitude of the early component of the contingent negative variation (CNV). These results confirm the interpretation of that wave as an orienting response. In contrast, the duration of the warning signal did not affect the second CNV component or reaction time latency. Additional dissociation between the two CNV components was evident in their distribution on the scalp. The early component was smallest at Pz, whereas the late component attained its lowest amplitude at Fz. Concurrently recorded palmar skin potentials exhibited different polarity and latency from the CNV.

1, Introduction Recent evidence derived from reaction-time foreperiods lasting several seconds has indicated that the contingent negative variation (CNV) is comprised of two negative waves (Loveless and Sanford, 1973, 1974; Weerts and Lang, 1973; Kiorman, 1975). The first response grows out of the evoked response to the warning stimulus and peaks at around 1 set post-onset whereas the second wave reaches its maximum level at the onset of the imperative stimulus. Weerts and Lang (1973) demonstrated that the first wave undergoes response decrement over trial blocks. In contrast, the second wave exhibits enhancement over time (Weerts and Lang, 1973; Klorman, 1975). Both Weerts and Lang (1973) and Loveless and Sanford (1974,1975) have proposed that the first wave represents an orientation reaction (OR) and the second wave an expectancy response. The interpretation of the first wave as an OR is buttressed by the emergence *Address for correspondence: Rafael Klorman, Department Rochester, Rochester, New York 14627, U.S.A. 263

of Psychology,

University of

ofsuch a wave in response to nonsignal tones (Jarvilehto and Fruhstorfer, 1973; Loveless, in press). In addition, Loveless and Sanford ( 1975) have reported an increase in amplitude of the first wave as a function of the physical intensity of auditory warning signals. The rationale of their approach is that if the early component of the CNV represents an OR, lawful changes in this response should be evoked by manipulation of the physical characteristics of the warning signal. In the present research, the temporal duration of the warning signal was varied. Graham (1973) has reviewed evidence relating this parameter of stimulation to the magnitude of physiological responses. Consequently, lawful changes in the early CNV wave were expected as a function of variations in the duration of the warning signal. An additional approach to the study of differences in the psychological correlates of the two components of the CNV involves the examination of their distribution on the scalp. Loveless (in press) has shown that the first CNV wave has a frontal-dominant distribution, whereas the terminal component is criteria assumed to peak at the vertex (Cohen, 1969). Thus, topographical might provide further evidence on dissociation of the two components of the CNV. Therefore, in the present research, EEG data were collected from three different scalp sites along the midline. A final focus of the present research concerned the presence of electrodermal contaminants in the CNV (Corby. Roth and Kopell, 1974). To this end, concurrent recordings of palmar skin potential were obtained. These data were collected to assess the morphological and temporal covariation between palmar and scalp electrical activity. 2. Method The subjects were 24 Caucasian male undergraduates who fulfilled part of a course requirement for participation in research. All reported an unremarkable health history and denied current illness or use of drugs. As a partial control for circadian rhythms, all subjects were tested in the afternoon.

Chlorided silver Grass EEG electrodes were attached with collodion to three scalp sites (Fz, Cz and Pz) and referenced to two linked chlorided silver earclips. In addition, Beckman miniature electrodes were taped to the center of the forehead (ground) and the left supra- and infraorbital ridges to detect vertical EOG. All sites were abraded with Cambridge jelly so as to bring impedances below 3 k0, as measured by a Grass EZM meter. For I8 sub.jects, skin potential was monitored from a Beckman miniature electrode attached to the medial phalanx of the index linger on the non-dominant hand and referenced to a comparable electrode placed on a drilled site on the corresponding forearm. An isotonic electrolyte (Venables and Christie, 1973) was employed for this measure. All signals were amplified with direct coupling by a Beckman

Warning-signal duration, early and late CN V components

type R dynograph and recorded subsequent computer analysis.

on a Vetter model A FM tape recorder

265

for

2.2. Tusk The warning signal was a tone of 750 Hz at 70 dB delivered coincidently with illumination of a red neon light that served as fixation point. Offset of the light was the imperative signal. This paradigm, which involved coincident onset of warning signal and fixation point coupled with termination of the latter as imperative stimulus, was recommended by Weerts and Lang (1973). These authors noted that ocular fixation reduces CNV amplitude unless the fixation point also serves as imperative stimulus. The two warning-stimulus durations employed were 0.5 and 2 sec. Insofar as use of a longer warning signal necessarily extends the foreperiod, foreperiod length was also controlled. Therefore, four conditions were formed which involved combinations of warning tone duration and the segment of the foreperiod elapsing between offset of the tone and that of the fixation light. The resulting four paradigms were as follows: (a) 0.5-5.5: the warning tone lasted 0.5 set and the light outlasted it by 5.5 set; (b) 2-5.5: warning-tone duration was 2 set, and the light was extinguished 5.5 set after tone offset; (c) 2-4: the warning tone was on for 2 set and the light for 4 set following termination of the tone; and (d) 0.5-4: the warning tone lasted 0.5 set and the light outlasted it by 4 sec. It should be noted that these paradigms control both total foreperiod duration (onset to offset of fixation light) and the segment of the foreperiod elapsing between the end of the warning signal and that of the foreperiod. Each condition was administered in a block of 20 identical trials. The experiment was designed according to a latin square involving four orders of conditions, to each of which six subjects were randomly assigned. 2.3. Procedure The session took place in an IAC chamber. Mean temperature was 22.22”C and mean humidity was 47.67 y/;. The subject was seated in a lounge chair and the lights were dimmed. Following a 15-min adaptation period, the task was described to the subject. He was instructed that, upon offset of the fixation light, he should press a microswitch held in his dominant hand as quickly as possible while avoiding premature responses. Three to five practice trials were administered in order to familiarize the subject with the assigned condition and to provide practice with ocular fixation. Subsequently, the appropriate condition was administered. Intertrial intervals ranged from IO to 30 sec. Following a 2-min break separating conditions, the procedure was repeated. Lehigh Valley solid state logic was employed to program stimulus presentation and to measure response latencies.

266

R. Klornmn

and E. Bentsen

2.4. Data processing EEG and EOG data were analyzed offline by a DEC PDP-12 computer. The EEG data were passed through a Krohn-Hite model 330 filter, which eliminated low-frequency drift below 0.01 Hz (Weerts and Lang, 1973); this filter setting is equivalent to a time constant of approximately 15.92 sec. A sampling rate of 20 Hz was employed by the computer for an epoch spanning 2 set preceding and 10.5 set following each warning signal. From the 20 trials administered in each condition, the last 10 trials free of ocular artifact were retained for averaging. The average for each condition was smoothed such that each point was replaced by an average consisting of that value as well as the adjacent pair of points. Subsequently, a baseline derived from the three intervals (150 msec) preceding the onset of each condition was subtracted from each data point in the corresponding average. A computer program identified several peaks of each curve based on rules derived from earlier work (Klorman, 1975). Description of these parameters is facilitated by reference to the averages of an illustrative subject, which are displayed in fig. 1. The program identified the peak negativity that occurred following the positive trough of the evoked response and within 2 set postonset. Previous work (Klorman, 1975) indicated that the early negative wave reached its peak at around 1 set post-stimulus, decayed slowly for approximately 1 set, and then decayed more steeply until reaching a nadir at around 3-3.5 set post-stimulus. An estimate of the slope of the initial gradual decay was derived as the difference between the value at 1 set post-onset (the mean of the intervals representing 0.95, 1.OO, and 1.05 set post-stimulus) and that at 2 set post-warning signal (the mean of the values sampled at 1.95, 2.00, and

-24:

’

X--x

I

2-5.5 24,

=I

+zY4 0

2

._ ~__ 4 6

SECONDS

8

cI_.

0

2

4

6’8

SECONDS

Fig. 1. Vertex potentials of a single subject averaged separately over each condition. Upward deflections represent relative negativity at the vertex. Conditions are coded by the duration of the warning signal and the interval elapsing between the offset of the warning tone and the end of the foreperiod.

Warning-signal

267

duration, early and late CNV components

2.05 set post-onset). The terminal voltage of the average was defined as the average of the last three values (150 msec) in the foreperiod. Another estimate of the late component of the curve was obtained by subtracting from the terminal voltage the least negative point in the already noted trough occurring between 3 and 3.5 set post-onset. Skin potential activity was scored manually from the polygraph records. The main purpose of these analyses was to examine potential electrodermal artifacts in the CNV. It was apparent that the study of skin potential activity in its own right was impeded by the brevity of the intertrial intervals and the short duration of the foreperiod in the 0.5-4 condition (cf. Stern, 1972). Therefore, no attempt was made to perform a separate assessment of mono-, di- or triphasic responses (Edelberg, 1972). The usual latency of electrodermal activity (Edelberg, 1972) suggests that responses with latency under 1 set may not be attributed to stimulation. Therefore, the foreperiod for each condition was divided into intervals the end points of which were 1,3,4.5,6, and 7.5 sec. Electrodermal responses were scored from peak to peak and coded as to the segment of the interval in which they began. This procedure accords with conventions for scoring electrodermal data (Prokasy and Kumpfer, 1973) but does not yield temporal averages analogous to the CNV. The amplitude of the last response in the foreperiod was taken as its amplitude at the offset of the fixation light. However, scoring of the peak of that response often required following it into the intertrial interval. Electrodermal activity occurring after the foreperiod’s end (4.5, 6, or 7.5 set) was not scored because concurrent EEG activity was contaminated by ocular movement.

Amplitudes

of components

Table 1. of the CNV wave (all measures are tabulated in microvolts). Measure

Lead

Fz

cz

Pz

Condition

Early wave

Terminal level

Late wave

2-5.5 0.54 2-6 0.5-5.5

-13.52 -11.03 -12.85 -10.61

-1.33 -0.66 -1.58 -0.18

-1.90 -3.06 -2.05 -3.21

225.5 0.54 2-6 0.5-5.5

-13.64 -11.10 -13.51 -11.75

-3.41 -4.48 -3.34 -3.83

-2.83 -4.33 -2.51 -4.11

2-5.5 0.54 2-6 0.5-5.5

-9.05

-4.18 -5.24 -4.54 -4.18

-3.43 -4.34 -3.53 -3.11

-8.84 -8.83 -8.38

3. Results Table I displays the mean over each condition for all EEG measures. Each measure was subjected to a replicated latin square analysis of variance followed by planned comparisons of: (a) the average of conditions with a short warning signal versus those with a long warning signal; (b) the two conditions with short warning signals; and (c) the two conditions with long warning signals. It was predicted that only the contrast testing the effect of warning-signal duration would attain significance.

Figure 2 displays the vertex-EEG curves averaged over all subjects for each condition. All curves contain an early negative wave that peaks around I .5 set post-onset and a second wave that peaks preceding the end of the foreperiod. However, the curve for the 2--5.5 condition exhibits a slight decline in the last 0.5 set of the interval. A similar morphology characterized frontal and parietal data. By far the most noteworthy aspect of fig. 2 is that the two conditions involving the 2-set warning tone elicited early waves of greater amplitude whereas no such effects were apparent on the second CNV component. Statistical analyses of vertex data confirmed these impressions. The amplitude of the early CNV on the two conditions with long warning signals exceeded the magnitude of that wave on trials initiated by short warning tones (F( I /60) = 6.3 I. p c 0.025). whereas neither contrast involving conditions with like tone duration achieved significance (both F (I/60) < I). Similarly, for the Fz-derived early CNV, the comparison testing warning-signal duration was significant (F (l/60) = 6.36. /J < 0.025). whereas the remaining comparisons were not (both F (I/60) < I). However, for the parietally derived early CNV, none of these comparisons achieved significance (all F (l/60) < I).

-12, - 8j

I

07

4

~6

8

SECQMDS

Fig. 2. Vertex potentials

SECONDS

averaged

over all subjects

for each condition.

Warning-signal EARLY

duration, early and late CNV components WAVE

TERMINAL

LEVEL

LATE

269

WAVE

Fig. 3. Distribution over the three recording sites on the scalp of three measures of the CNV, averaged over conditions.

Figure 3 depicts each component of the CNV as a function of lead derivation. The amplitude of the early CNV component was considerably smaller at the parietal than vertex (F (l/40) = 20.69, p < 0.001) or frontal (F (~/40) = 15.54, p < O.OOl) derivations. Data from the latter two leads did not differ from each other (F (l/40) < 1, NS). No effects of order or trial blocks were present in these data with one exception. The early component of the parietal CNV showed enhancement over trial blocks (trial blocks linear F (l/20) = 5.86, p < 0.025). The preceding analyses have documented augmentation of the amplitude of the early CNV at the more frontal derivations as a function of warning signal duration. Subsidiary analyses were performed as a check on the possibility that these differences might be due to differential latencies of decay of the early wave. Because of the slow rise and decay of these waves, manual scoring of the slope of decay was not attempted, and this measure was estimated, as described in section 2.5, by the drop in voltage occurring between second 1 and second 2. Statistical analysis indicated that the decay slopes of the early CNV waves did not differ over conditions for frontal, vertex, and parietal dat,a (F (3/60) = 1.55, 1.46, and < 1, NS, respectively). 3.2. Lurecfvv wai’e As noted earlier, the late component of the CNV was scored as both the terminal level of the wave and as the difference between the latter value and the trough occurring in the middle ofthe interval. No significant differences among conditions emerged for either measure for any derivation. Once again, order did not achieve significance in any analysis. The only findings on temporal trends involved the marginal tendency of the Cz- (trial blocks linear F( l/20) = 3.65, p < 0.10) and Fz-derived (F (l/20) = 3.51, p < 0.10) terminal levels to display enhancement over the session. As shown in fig. 3, the topography of the later component of the CNV was the reverse of that obtained for the early wave. A significantly smaller amplitude characterized the frontally derived terminal level relative to the compar-

210

R. Klovman and E. Bentsen

able measures obtained from the vertex (F( l/40) = 6.71, p < 0.025) or parietal (F(l/40) = 10.89,~ < 0.005) leads. The latter two leads did not differ from each other (F( l/40) < I). Analyses of the second wave, when scored relative to the mid-foreperiod trough, disclosed similar but non-significant trends. The amplitude of this measure tended to be slightly greater at the parietal than at the frontal lead (F(l/40) = 3.35, p < 0.10). 3.3. Skin potentiul responses Table 2 displays several measures of electrodermal activity and mean reaction time for each condition. For each segment of the foreperiod, the table presents the mean amplitude over trials included in CNV averages and the proportion of such trials on which a response limb began in that part of the interval. The latency of the peak of the last response in the foreperiod is based on the mean for trials on which activity was present and excludes subjects who exhibited zero responses. Skin potential activity in the first second of the foreperiod was present on about half of the trials. Activity in the first second was characterized by a positive polarity and on 78 9,; of the trials on which electrodermal change was detectable, it began prior to the onset of the warning signal. Thus, considerations of polarity and latency of onset argue against electrodermal contamination of the early CNV wave. The only difference between conditions involved the smaller amplitude of deflections in the 0.5-5.5 than in the 225.5 treatment (F (l/51) = 5.90, p < 0.025). A negative response originated at an average of 2.20 set into the foreperiod, a point at which the CNV was in a positive phase. As was the case in the previous segment, this activity displayed smaller amplitude in the 0.5-5.5 than in the 2-5.5 condition (F ( l/S I) = 7.24, p < 0.025). Skin potential changes with onset in the third segment of the foreperiod began at an average of 3.76 set, were typically positive, and did not vary in amplitude over conditions (F(3/5l) = 1.32, NS). It is noteworthy that in the 0.5-4 condition the average electrodermal activity in this portion of the foreperiod exhibited the opposite polarity to the CNV and, unlike the late brain waves, peaked several seconds after the end of the foreperiod. Electrodermal activity commencing in the subsequent segment was characterized by low amplitude, doubtless because of the low number of subjects exhibiting a response. There were no differences in response amplitude between conditions (F(2/34) < 1, NS). Again, it is notable that in the two conditions with foreperiods terminating in this segment, the peak of electrodermal activity outlasted by several seconds the end of the foreperiod. Similarly, a low amplitude and absence of time-locking to the imperative stimulus characterized skin potential responses originating in the last segment of the 7.5~set condition.

0.06

0.02

2-4

0.5-5.5

0.41

0.49

0.52

0.49

Ph

-0.12

-0.32

-0.35 --

-0.46

2?

0.57

0.72

0.69

0.66

Pb -.

1.01-3

0.08

0.G

0.09

0.14

Xa

0.45

0.47

0.45

0.47

Pb --_

3.01A5

--

-0.02 ~-0.19

0.24

_e

_e 0.00

0.22

Pb

0.00

Xa

4.51-6

____-

9.61=

10.88’

_e ~_

_e

.._e

_e

_._~~~

9.739

10.56r

_e

0.11

Pa

Latency of last response’ --_ ..-.

_.e

0.00

x3

6.01-7.5

“In millivolts; bproportion of trials on which a response occurred; c in seconds; “in milliseconds; %ot scored; ‘PI= 9; and en = 16.

0.05

0.08

2-5.5

0.5-4

Xa

Condition

o-1

Segments of the foreperiod

Table 2. Skin potential and reaction time measures.

314

318

311

305

Reaction time

_~

2z z

a

2 -Y :!

f a s. “a 2 G % c. 6 P

2

2

g.

212

R. Klovtnan and E. Bentsen

3.4. Electra-oculogruphic actidty The EOG potentials elapsing between the onset and end of the foreperiod were scored separately for each condition. These data averaged 13.29 & 2.41 I_IV when measured without regard to polarity. When direction of eyeball movement was taken into account, EOG averaged 2.16 + 3.38 uV in the downward direction. In addition to its moderate amplitude, EOG activity displayed flat records that did not resemble the diphasic CNV waveform. 3.5. Response lutenq~ Comparable reaction times were obtained NS).

in the four conditions

(F(3/60)

< I,

4. Discussion The present results provide additional evidence for the existence of early and late CNV components. As noted by other investigators (Loveless and Sanford, 1974; Weerts and Lang, 1973), the two waves seem to reflect different psychological processes, which may be confounded in the study of the CNV in short foreperiods. The present results indicated that the early wave meets a property of the OR, insofar as its amplitude is augmented by warning-signal duration. In combination with prior data on the effects of the physical intensity of the warning signal (Loveless and Sanford, 1975), the results lend credence to the interpretation of the early wave as an OR. In fact, the absence of such a phenomenon in the late CNV component supports the view that the two CNV waves reflect different psychological processes. In a different context, Keidel (1971) has reported that the duration of nonsignal auditory cues coincides closely with the duration of the resulting slow EEG waveforms, which are comparable to our early waves. This timelocking appears especially striking with tones longer than I set, the duration of which exceeds that of the evoked response to onset of stimulation. The duration of the early waveforms evoked by the 2-set signal in the present research is consistent with this observation. This finding offers further support to the interpretation of the early wave as an orientation reaction. The finding that the late wave was unaffected by foreperiod duration may appear puzzling in view of reports (McAdam, Knott and Rebert. 1969; Loveless and Sanford, 1975) that longer foreperiod durations reduce the terminal level of the CNV. However, those reports are based on a range of foreperiod durations (l-8 set) including much shorter intervals than those employed in the present study (4.5-7.5 set). It is likely, then, that the foreperiod durations sampled in the present study exceed some critical value beyond which additional attenuation does not occur. Similarly, the relatively narrow range of foreperiods employed may account for the lack of differential reaction time latencies as a function of foreperiod length. Warning-signal duration also did not affect reation times, probably because this experimental manipulation

Warning-signal duration, early and late CNV components

273

concerned the early part of the foreperiod. Loveless and Sanford’s (1975) manipulation of warning-signal intensity augmented early CNV amplitude and reduced reaction time, but in that work the auditory signal was presented throughout the foreperiod. In the present study, neither behavioral nor electrophysiological measures derived in the latter part of the foreperiod varied across experimental conditions. Previous findings concerning response decrement of the early component of the CNV and enhancement of the late wave over trial blocks were not replicated. In fact, the early wave of the parietal CNV exhibited increment over time. It should be noted that the design of the present study, which involved alternating durations of warning signal, militated against habituation. Previous studies demonstrating temporal decrement of the early wave (Weerts and Lang, 1973) held constant throughout the experiment the physical characteristics of the warning signal. On the other hand, additional research is required to determine the reasons for the marginal support obtained for previous reports (Weerts and Lang, 1973; Klorman, 1975) of enhancement of the late wave’s amplitude over trial blocks. In addition to the already noted divergence in the effects of warning-signal duration on the two CNV waves, further evidence of dissociation is provided by their differential topographic distribution. The present results replicated Loveless’ (in press) report of a frontal-dominant distribution for the early wave. However, in Loveless’ report, frontal waves exceeded those derived from the vertex lead. In the present data, in contrast, the early wave was non-significantly smaller at the frontal than vertex sites. The two studies differed, however, in that a motor response was not required in Loveless’ investigation. The more posterior peak of the late wave is in accord with the distribution of the readiness potential (Deeke, Becker, Grozinger, Scheid and Kornhuber, 1973). This interpretation of the late wave would be reinforced by evidence of laterality in its distribution Additional research is required to resolve these issues. The findings on skin potential data bear a relevance to Corby et al.‘s (1974) work on cephalic electrodermal artifact in skin potential. Admittedly, the present electrodermal data cannot provide a definitive answer to this issue because of their non-cephalic source. It should be noted that Picton and Hillyard (1972) found that ‘palmar potential changes were frequently observed in the absence of any cephalic deflections, and occasionally cite z.ersa (p. 421, italics in the original)‘. Perhaps, then, palmar electrodermal activity may represent an upper limit of cephalic skin potential changes. In any event, the palmar data did not suggest any evidence for such an artifact. Electrodermal activity in the second during which the early negative component of the CNV developed was characterized by a positive polarity and its onset typically preceded the warning signal. In addition, the duration of the warning tone did not affect skin potential amplitude in a systematic way, as was the case with the early CNV wave. Finally, skin potential activity in the latter portion of the

274

R. Klorttmtt nntl E. Bm!scvr

foreperiod was rather infrequent and failed to exhibit time-locking with respect to the imperative event. It is noteworthy in this context that Lacey and Lacey’s (1973) concurrent averages of skin potential and CNV were also discordant with regard to polarity. The failure ofwarning-stimulus duration to enhance electrodermal response amplitude should not be viewed as contradicting prior work on this phenomenon in autonomic responsiveness (Graham, 1973). Nor does the lack of an anticipatory skin potential response indicate an ineffective conditioning paradigm. As noted earlier, the present experimental conditions were not propitious for the study of electrodermal activity in its own right. The present investigation employed a paradigm consisting of a mixture of trace and delayed conditioning procedures, since warning and imperative signals overlapped over time. Other workers have employed trace (Loveless and Sanford, 1973 ; Weerts and Lang, 1973) and delayed paradigms (Loveless and Sanford, 1975). While the latter procedures evoked less clear differentiation of the two CNV components, diphasic waves were obtained with either paradigm. These studies, then, indicate that the diphasic CNV waveform is not specific to a particular reaction-time procedure. However, only auditory warning signals have been employed thus far. Future research is required to establish whether other modalities elicit comparable CNV waveforms. Although this and other questions remain, the present results, in combination with prior investigations, reinforce the interpretation of the first CNV component as an orienting response and the existence of multiple components in slow brain potentials. Acknowledgements Support for the second author was provided by a stipend from NIMH grant MH05 I46 and computer time was funded by a special grant from the University of Rochester College of Arts and Science. We are grateful to Kenneth Ouriel for aid in scoring records and to Dale McAdam, Lecnard Salzman, Jerome Schwartzbaum, and Alan Wiesenfeld for facilitating the research. References Cohen, J. (1969). Very slow brain potentials relating toexpectancy: the CNV. In: Donchin, D. and Lindsley, D. 13. (Eds.) nr.r~u~c El~ohctl fotottin!s. National Aeronautics and Space Administration: Washington, D.C., 143--198. Corby, J. C.. Roth, N. T. and Kopell, B. S. (1973). Prevalence and methods of controi of the cephalic skin potential EEG artifact. PsJ~c~hot~/t~~~\ioiot,pI~, Il. 350-360. Deece, L., Beck&, W., Grozinger, B., Scl;eid,‘P..and Kornhuber, H. (1973). Human brain potentials preceding voluntary limb movements. Elc~c/t~o~ttcc~~~~t~tl~t~t~uplt~ cm/ Clinical Netrr.opll)siolog~~, Supplement 33, 87-93. Edelberg, R. (1972). Electrical activity of the skin. In: Grecnfield, N. S. and Sternbach, R. A. (Eds.) Hutd/~ook of’f.s~c/tctplr~.rio/~t~~~. Holt, Rinehart and Winston: New York, 367-418. Graham, F. K. (1973). Habituation and dishabituation of responses innervated by the autonomic nervous system. In: Peeke, H. V. S. and Herz, M. J. (Eds.) Htrhitrtntion: Bdzacioral Studies and Pltysiological Sttixtrates. Academic Press : New York, 163-2 18.

Warning-signal

duration,

early and late CNV components

275

Jarvilehto, T. and Fruhstorfer, H. (1973). Is the sound-evoked DC potential a contingent negative variation ? Electroencephalography and Clinical Nearophysiology, Supplement 33, 105~108. Keidel, W. D. (1971). D.C.-potentials in the auditory evoked response in man. Acta Ofnlaryngologica, 71, 242-248. Klorman, R. (1975). Contingent negative variation and cardiac deceleration in a long preparatory interval: a developmental study. Psychophysiology, 12, 609-617. Lacey, J. 1. and Lacey, B. C. (1973). Experimental association and dissociation of phasic bradycardia and vertex-negative waves: a psychophysiological study of attention and response intention. Electroencephalograpl~y and Clinical Neurophysiolog~~, Supplement 33, 281-285. Loveless, N. E. (In press). Distribution of response to non-signal stimuli. In: McCallum, W. C. and Knott, J. R. (Ed?.) Third International Congress on Ewnt RelatedSlow Potentials of the Braiil. Loveless, N. E. and Sanford, A. J. (1973). The CNV baseline: considerations of internal consistency of data. Electroencephalography and CliXcal Nerwophysiology, Supplement 33, 19-23. Loveless, N. E. and Sanford, A. J. (1974). Effects of age on the contingent negative variation and preparatory set in a reaction-time task. Journalof Gerontology, 29, 52-63. Loveless, N. E. and Sanford, A. J. (1975). The impact of warning signal intensity on reaction time and components of the contingent negative variation. Biological Psvchology, 2, 2 17-226. McAdam, D. W., Knott, J. R. and Rebert, C. S. (1969). Cortical slow potential changes in man related to interstimulus interval and to pre-trial prediction of interstimulus interval. Psychaph~,siof~)g.v, 5, 349-358. Picton and Hillyard (1972). Cephalic skin potentials in electroencephalography. Electroencephalography and Clinical Newophysiology, 33, 419-424. Prokasy, W. F. and Kumpfer, K. L. (1973). Classical conditioning. In: Prokasy, W. F. and Raskin, D. C. (Eds.) Electrodwmal Activity in P.svcholopical Research. Academic Press: New York, 157-201. Stern, J. A. (1972). Physiological response measures during classical conditioning. In: Greenfield, N. S. and Sternbach, R. A. (Eds.) Handbook of P.yychophvsio/ogy. Holt, Rinehart and Winston: New York, 197-227. Venables, P. H. and Christie, M. J. (1973). Mechanisms, instrumentation, recording techniques, and quantification of responses. In: Prokasy, W. F. and Raskin, D. C. (Eds.) Electrodermal Actil?ty in Psychological Research. Academic Press: New York, l-124. Weerts, T. C. and Lang, P. J. (1973). The eRects of eye fixation and stimulus and response location on the contingent negative variation (CNV). BiologicalPsychology, 1, l-19.