Habituation of P300 and reflex motor (startle blink) responses to repetitive startling stimuli in children

INTERNATIONAL JOURNAL OF PSYCHOPHYSIOLOGY

ELSEVIER

International

Journal of Psychophysiology

22 (1996) 97-109

Habituation of P300 and reflex motor ( startle blink) responses to repetitive startling stimuli in children Chiaki Hirano a, Andrew T. Russell b, Edward M. Ornitz b7c7*,Minzhi Liu b7d a Nagahama Red Cross Hospital, Nagahama, Japan b Department of Psychiatry, University of California at Los Angeles (UCLA), Los Angeles, CA 90024, USA ’ Brain Research Institute. University of California at Los Angeles (UCLA), Los Angeles, CA 90024, USA ’ Department of Biostatistics. University of California at Los Angeles (UCU), Los Angeles, CA 90024, USA Received

17 October

1995; revised 12 January

1996; accepted 23 January

1996

Abstract Positive EEG deflections with the latency and scalp distribution of the P300 accompany startle in response to loud auditory stimuli in a non-task context. The purpose of this investigation is to determine if habituation would have effects on the P300 similar to those on the startle blink. Thirty-four normal 7 to 1 l-year-old boys from a startle habituation study had EEG recordings of sufficient quality to provide data for the current study. Startle was measured both as orbicularis oculi EMG and vertical EGG and P300 was recorded at Pz in response to 40 104 dB bursts of white noise presented at 23-s intervals. Both the startle response and the P300 habituated toward asymptotic levels after the first 28 trials, suggesting that both startle and the subsequent cognitive evaluation of the startling stimulus, reflected in the P300 response, are modulated by a common neurophysiological mechanism extrinsic to the direct startle pathway. A modest significant correlation between the P300 and the vertical EGG peak latencies for the initial trials suggests that the cognitive evaluation of the startling stimulus may also include evaluation of the reflex response to that stimulus. Analyses of the within-subject associations between startle and P300 initial amplitudes and rates of habituation showed that these parameters varied independently within the individual subject, suggesting that the P300 is not a component of the startle response. Rather, it reflects an evaluation of the startling stimulus, decreasing in amplitude as the surprising value of the startling stimulus decreases with habituation. Keywords:

Habituation;

P300, Startle; Blink; Children

1. Introduction

Event-related potentials panying startle responses

(ERP) in the EEG accomwere originally described

* Corresponding author. 27-384C NPI, UCLA, 760 Westwood Plaza, Los Angeles, CA 90024, USA. Tel.: (310) 8256025; fax: (310) 82.5-2982. 0167-8760/96/$15.00 0 1996 Elsevier Science B.V. All rights reserved PI/ SO167-8760(96>00018-9

by Larson (1956, 1960a, b), Sakano and Pickenhain (19681, and Davis and Heninger (1972). Recently, interest has focused on the association of the P300 component of the ERP when ERPs are elicited automatically in association with startle responses to intense stimuli (Putnam and Roth, 1987, 1990; Ford and Pfefferbaum, 1991; Ford et al., 1994; Sugawara et al., 1994).

98

C. Hiratw et al./lntermtional

Journal

1.1. Effects of stimulus conditions and experimental context on cognitive and automatic P300 The P300 component can be considered a cognitive response to an attended low-probability task-relevant stimulus in that it reflects the mental evaluation of surprising events, surprising in the sense of being ‘unexpected’ but relevant (Donchin, 1981), and hence reflects mental revisions of the subject’s current model of the environment (referred to as ‘context updating’, Donchin and Coles, 1988). However, it can occur in ‘passive’ or ‘automatic’ experimental contexts in which no task is assigned and no motor response is required. It has been generated passively or automatically in response to low-probability stimuli in experiments in which the subjects are uninstructed (Ford et al., 1994), engaged in preoccupying or distracting activities, e.g., working puzzles, daydreaming, or reading (Ritter et al., 1968; Squires et al., 1975; Polich, 1987, 1989a), or even under anesthesia (Plourde et al., 1993) or comatose (Yingling et al., 1990). The habituation of the P300 response to low-probability stimuli, designated explicitly or implicitly as ‘targets’, has been studied in a series of recent experiments (Polich, 1987, 1989b; Lew and Polich, 1993; Polich and McIssac, 1994). Target-generated P300 proved to be very resistant to habituation across the first 15 to 20 trials, even under relatively ‘passive’ conditions (Polich and McIssac, 1994), with significant habituation occurring only after 40 to 60 trials (Lammers and Badia, 1989; Polich, 1989b; Lew and Polich, 1993). After about 60 trials, habituation of target-generated P300 does occur, depending on inter-target interval and inter-trial-block interval (Lammers and Badia, 1989; Wesensten et al., 1990). The P300 components in all of these studies were generated in response to auditory stimuli of mild to moderate intensity (usually 60 dB SPL) and did not show amplitude differences in response to sound intensity levels varying between 30 and 70 dB SPL (Polich, 1989~). However, when sound intensities ranged between 65 and 110 dB SPL (levels associated with evoked startle blinks), then P300 amplitudes increased with increasing sound levels (Roth et al., 1984). In most P300 paradigms, the ‘target’ stimulus to which the subject may be instructed to respond

of Psychophysiology

22 (1996) 97-109

(active or effortful paradigm) or not (passive or automatic paradigm) is contrasted with a ‘standard’ stimulus. This is the ‘oddball’ task. However, it has been shown that an equivalent P300 response can be elicited in a single-stimulus paradigm in which the only stimulus is the ‘target’ stimulus (Polich et al., 1994). In summary, while the P300 can be considered a cognitive response, indexing stimulus evaluation and information processing (Johnson, 19881, it can be elicited passively and automatically, independent of a motor or non-motor response, and in response to only one type of stimulus presented repetitively at relatively long intervals. It is resistant to habituation, and it is responsive to the intensity of the stimulus, when the range of stimulus intensities is high enough to generate startle responses. I .2. P300 evoked by startling stimuli What then is the nature of the relationship of the P300 response to a motor response, i.e. a startle blink, when the latter is evoked by the same stimulus? Using the vertical EOG (a measure that will miss small reflex blinks detected by orbicularis oculi EMG recording), correlations between startle blink amplitudes and P300 amplitudes were not significant and P300 amplitudes were similar whether accompanied by a startle blink or not in young adults (Ford and Pfefferbaum, 1991). However, across an age span including young and middle-aged adults, P300 amplitudes were significantly correlated with the number of startle blinks elicited by startling stimuli (Ford et al., 1994). Using the orbicularis oculi EMG as the measure of startle blinks, Putnam and Roth (1990) found that the P300 failed to show the sensitivity to stimulus rise time that is characteristic of startle blinks. Sugawara et al. (1994) found that the P300 elicited by startling stimuli was modulated by prestimulation in the same manner as the startle response itself, i.e. showing inhibition when the non-startling prestimulus preceded the startling stimulus by a short interval and facilitation following a long prestimulation interval (Graham, 1975; Ornitz et al., 1986). I .3. P300 and startle habituation The relationship of P300 response and startle response changes during habituation requires further

C. Hiram

et al./lnternational

Journal

study. Putnam and Roth (1990) stated that ‘P300 habituated more slowly than’ startle without providing direct statistical evaluation of the suggested difference in habituation rates. Perusal of their data suggests that habituation of both P300 and startle reached asymptotic levels after about the same number of trial-blocks. In a startle habituation paradigm in which the stimuli evoking startle are not designated as targets and are separated by relatively long interstimulus intervals, the P300 component of the accompanying ERPs is elicited under conditions that approximate P300 generation during passive, automatic, and single-stimulus conditions but do not generate motor responses. In the latter cases, the P300 has been resistant to habituation (Polich, 1989b; Lew and Polich, 1993; Polich and McIssac, 1994). In contrast, when evoked by stronger stimuli that also evoke startle, the P300 does habituate (Putnam and Roth, 1990). This suggests that the presence of a motor response, the startle blink, and its habituation, or the neurophysiological mechanism underlying startle blink habituation, somehow influences the habituation of the P300 response. This type of association occurs in respect to prestimulation modulation of startle (Sugawara et al., 1994) where the underlying functional neuroanatomy (a pathway in the mesopontine brainstem tegmentum) is fairly well delineated (Leitner et al., 1981; Leitner and Cohen, 1985; Saitoh et al., 1987; Wu et al., 1988). For startle habituation, the underlying neurophysiological mechanism is not so well defined. The purpose of this investigation is to evaluate the associations between or the independence of P300 and startle during habituation, using as rigorous statistical procedures as possible, in an experimental procedure that incorporates certain features of the passive and automatic and single-stimulus P300 paradigms. Hence, only one type of stimulus, the startling stimulus, was given, at relatively long intervals (23 s), with neither task nor instruction. Additionally, to enhance the passive and automatic aspects of the stimulus-response relationship, subjects watched a silent movie or cartoon of interest while receiving the startling stimuli. The subjects were normal school-age boys drawn from a larger investigation of the development of startle habituation throughout childhood. Two statistical approaches were employed to study relationships between startle

of Psychophysiology

22 (1996) 97-109

99

and P300 habituation. A within-subject procedure examined the intrasubject covariance between the linear component of habituation of the two variables. An across-subject procedure compared the timecourse of the habituation of the two variables by evaluating the time of development of non-linear components of habituation.

2. Materials

and methods

2.1. Subjects Putatively normal 7 to 1 l-year-old boys were recruited from the general population. Each subject and at least one of his parents were interviewed with the Diagnostic Interview for Children and Adolescents (DICA-C and DICA-P; Welner et al., 1987) by the second author (ATR), a child psychiatrist. The WISC-R (Wechsler, 1974) was administered to each subject. A parent of each subject completed the Child Behavior Checklist (Achenbach and Edelbrock, 1983) and the Werry-Weiss-Peters Rating Scale (Routh et al., 1974). A teacher of each subject completed the Child Behavior Checklist (Achenbach and Edelbrock, 1986) and the ADHD Comprehensive Teacher Rating Scale (ACTeRS; Ullmann et al., 1984, 1985). Based on this diagnostic evaluation, exclusion criteria were a full-scale IQ < 85, evidence of hearing or visual impairment, or any medical, neurological, psychiatric, or developmental disorder including more than three ADHD symptoms out of the 14 symptoms listed in DSM-III-R (1987). Thirty-four normal 7 to 1 l-year-old boys (115.7 + 19.6 months old) were entered into the study and were successfully recorded in the startle habituation experiment. Since the main interest of this study is the relationship between startle habituation and P300 habituation, these 34 subjects were selected from a larger subject pool on the basis of two criteria. Subjects with startle blink responses on the first block of four trials (prior to habituation) that were close to the threshold of measurement were not included to minimize floor kffects in the habituation study. They had stable EEG recordings (free of movement-related artifact) at Pz during most of the startle trials to minimize the number of trials that could not be used in the data analyses.

C. Hiratw et al./lnternational

100

Journal of Psychophysiology 22 (1996) 97-109

2.2. Stimuli Forty startle stimuli [SS, 104 dB (SPL), zero rise time, 50 ms white noise bursts] were presented every 23 to 25 s binaurally through TDH-49 circumaural earphones. The computer was programmed to present SS every 23 s but to wait up to but never longer than an additional 2.0 s while attempting to present the stimulus after 200 ms of tonic background orbicularis oculi EMG that did not exceed a predetermined maximum level. This compromise minimized the number of trials that were not usable, due to fluctuations in background EMG or spontaneous blinks occurring just before the stimulus and the reflex response. 2.3. Dependent

variables

Orbicularis oculi EMG was recorded bipolarly from 10 mm gold-plated cup electrodes taped to the skin approximately 11 to 12 mm apart, 9 mm below the left lower lid margin, with the lateral electrode 5 mm medial to the outer canthus. Measurements were confirmed after the electrodes were in place. The raw EMG was AC amplified, at a constant gain for all subjects, with filters set at half-amplitude 100-1000 Hz, then rectified and smoothed through a Coulboum contour-following integrator (time constant 3.5 ms), the output of which was digitized on a DEC 11/23 laboratory computer at a 500 Hz sampling rate. Onset latency was measured as the first increment greater than two standard deviations above the average baseline, computed for 200 ms before SS onset, that was not followed by a return to that level within a window 20-80 ms following SS onset. Peak amplitude was defined as the highest point within a window from response onset to 105 ms following SS onset, and was measured relative to a 120 ms pre-SS onset baseline. ’ These variables were used as the measures of the EMG startle response.

’ The peak amplitude of the orbicularis oculi EMG response (as well as the peak values of the EGG and the P300 described below) was measured on an averaging program that utilized a 120 ms pre-SS baseline. The onset latency of the EMG response was measured on a different program (see below) that utilized a 2C0-ms baseline.

The vertical electro-oculogram (EOG), recorded between silver-silver chloride electrodes placed just above and below the orbit in a vertical line through the pupil, was DC amplified. The output was digitized at a 250 Hz sampling rate. The peak amplitude and peak latency of the EOG were measured relative to a 120 ms pre-SS baseline within a window 20 to 150 ms following SS onset. These variables were used as the measures of the EOG startle response. The EEG, recorded between silver-silver chloride electrodes placed at Pz and the right mastoid, was AC amplified with filters set at half-amplitude O.l100 Hz. The output was digitized on the DEC 1 l/23 at a 500 Hz sampling rate. The peak amplitude and latency of P300 were measured at the most positive potential relative to a 120 ms pre-SS baseline within a window 250 to 380 ms post-SS onset.

2.4. Procedure Written informed consent was obtained from the parents of each child according to UCLA Human Subject Protection Committee approved procedures. Each subject had an adaptation session that took place between 7 and 3 days prior to the experimental session. The adaptation session included familiarization with laboratory personnel and the physical aspects of the laboratory, 20 min of watching silent TV cartoons or movies, and application of all electrodes and the earphones, and an audiometric test. A single SS was presented during the adaptation session. Hence, each subject was exposed to all aspects of the experimental situation except the repetitive presentation of the SS. The audiometric test was repeated just before the startle habituation session. Hearing in each ear was tested at 20, 30, 40 and 60 dB SPL at 1000, 2000, and 4000 Hz. All subjects entered into the experiment passed the audiometric test. To reduce boredom and maintain a constant level of alertness, and minimize gaze shifts and movement, subjects watched silent TV movies or cartoons of their choice while listening to the 40 SS. The first trial was presented after a 3 to 4 min adaptation period while the subject watched the silent movie. Subjects were asked to sit quietly and enjoy watching the silent TV movie or cartoon. There was no instruction either to pay attention to or ignore the

C. Hiram et al./lnternational

Journal of Psychophysiology 22 (1996) 97-109

sounds other than the fact that the sounds would be presented, repetitively. Subjects were observed continuously on a closed-circuit TV monitor focused on the subject’s face. EOG and EEG were monitored continuously from the DC vertical electro-oculogram and the EEG derivation respectively by visual inspection of a polygraph record. Drowsiness was defined as persistent lid closure and/or rolling eye movements (indicated by changes in the EOG recording) and/or persistent slowing and increased amplitude of the EEG. These physiologic indices were complemented by behavioral observations of the subject’s facial expression and eyes. 2.5. Data assessment (individual trials) The orbicularis oculi EMG, the vertical EOG and the EEG (Pz referred to the right mastoid) responses were assessed individually on each of the 40 trials from each of the 34 subjects prior to averaging. Trials with large DC shifts in the EOG or EEG or excessive muscle activity in the EMG during the 200 ms prior to SS onset were discarded. Trials were rejected prior to data analysis if activity in the EMG or EOG response indicated a blink onset prior to 20 ms after SS onset (on the assumption that such events would include spontaneous rather than reflex blinks). In the EEG channel, an N2 component was identified as the most negative potential between 160 and 290 ms post-SS onset. In the rare instance that an N2 [hence, a P2-N2-P3 (P300) complex] could not be identified, the trial was rejected. Although the N2 component could be recognized, it was not measured because the trailing voltage from the EOG recorded blink was great enough to contribute blink artifact to EEG values around 200 ms post-SS onset. Similarly, the early components (Pl, Nl , P2) of the event-related potential were not assessed because of likely contamination by the very large reflex blinks. The P3OO component was identified as the most positive potential following the N2 wave and occurring between 250 and 380 ms post-SS-onset. If activity in the EEG exceeded 150 PV at the latency of the P300, if activity in the EOG exceeded 100 PV at the same latency or if an obvious secondary blink was distinguished on the EOG channel within the window for P300, the trial was rejected.

101

Trials were rejected on-line if the polygraph record or observation of the subject on closed-circuit TV indicated movement or drowsiness at the time of the trial and off-line following data assessment as just described. An average of 4.91 trials per subject were rejected. 2.6. Averaging After individual trials were rejected, for each subject, consecutive blocks of four trials were averaged. A minimum of two trials of a given block had to be available for averaging from each subject. Hence, averaged evoked responses consisted of two to four trials. In the rare instance that either only one trial was present or an entire block of four trials was missing (five blocks in the total of 340 blocks in these data), the data for that block were imputed using the computer program BMDPSV (Dixon, 1992). An imputed value is the estimated conditional mean (based on the entire data set) of the missing response given the values of the responses that are present for the individual subject. On each averaged evoked response, the following measurements were made: the peak amplitude (relative to the 120 ms pre-SS baseline) and latency of P300 (after digital low-pass filtering at 35 Hz); the peak amplitude of the EMG, and the peak amplitude and latency of the EOG. Because the averaging program did not provide sufficient resolution, the onset latency of the EMG startle response was measured on individual trials, using a different program. These measurements were then averaged for each block. ’ 2.7. Data analysis The EMG startle response amplitudes were logtransformed to diminish the effect of skewness of their distribution. The distributions of P3OO ampli-

2Trials with zero EMG amplitudes were included. There were 29 such trials in the 1360 trials presented in these experiments. The onset latencies of these responses were replaced with the longest measurable latency of that subject’s trials, under the assumption that a very small response is probably present and, therefore, would have a latency. Longer latencies correlate significantly with smaller amplitudes (see Omitz et al. (1986) regarding the statistical basis of this substitution).

102

C. Hiram et al./Intermtiod

Journal

tudes, the EOG startle response amplitudes and all latencies did not require transformation. To study the habituation of the amplitudes and latencies of the two startle response measures (orbicularis oculi EMG and vertical DC EOG) and the P300, the 10 blocks of trials were analyzed with a repeated measures analysis of variance (Jennrich et factors. al., 1985) using blocks as intra-subject Greenhouse-Geisser epsilon (E) corrections were used for significance tests. Statistical analyses of associations between habituation rates and initial values both within and between variables were based on a random coefficient linear regression model (Laird and Ware, 1982; Bryk and Raudenbush, 1992) as implemented in the computer program BMDPSV. Random regression models are a special form of repeated measures analysis of variance with an intrasubject covariance structure based on the assumptions that the regression slopes do not have significant nonlinear trends and that the regression slopes and initial levels (the regression intercepts) vary randomly from subject to subject. Statistical inference about the mean slope and initial value is accomplished by using maximum likelihood methods to estimate the respective means and to compute standard errors and covariances of the estimates. Random regression methods may include several measures concurrently, making it possible to study the covariance between slopes of two variables within subjects. Maximum likelihood methods use Z-tests based on the asymptotic normality of the estimates. This process is described in detail in the documentation for BMDPSV. Associations between the startle response and P300 were based on the number of consecutive trial-blocks conforming to the assumption of no significant second or higher order trends as determined by repeated measures analyses of variance. A normal scores transformation was used to minimize the effect of single observation outliers in testing the covariance between slopes of the variables (Lehmann, 1975). Conover and Iman (1981) have noted the effectiveness of rank transformation in the analysis of repeated measures data. Normal score transformation is an extension of a simple rank transformation. In this process, each observed value in the data is replaced first by its rank within the full data set, then by the abscissa corresponding to the percentile

of Psychophysiology

22 (1996) 97-109

rank as it would appear in a normal distribution. This transformation appeared critical only when comparing variables; raw data values were analyzed in our study of habituation of individual variables. For those variables whose data did not conform to the assumptions underlying use of the random regression model, tests of association with startle amplitude were limited to simple correlations, using the mean values for the first block of 4 trials.

3. Results Generalized startle responses (observed on closed-circuit TV) occurred in a minority of the subjects and were limited to the first to fourth trial after which only the startle blink response was observed. As in previous studies, orbicularis oculi EMG startle blink amplitudes varied greatly in these normal subjects. To obtain a measure of this normal variation in response size amongst subjects, averaged EMG, EOG and P300 amplitudes from Block 1 were examined for each subject. Amongst subjects the averages of EMG startle amplitude ranged from 2.48 to 6.03 log-transformed A/D units. The EOG startle amplitudes ranged from 113.3 to 463.3 pV. P300 amplitudes ranged from -4.9 to 80.8 pV. Fig. 1 shows the grand averages of these measures across the 34 subjects for Block 1 and Block 7. The habituation of the two startle measures and P300 is indicated by the differences between Block 1 and Block 7. 3.1. Orbicularis

oculi EMG habituation

Fig. 2 (bottom left) shows the grand average of EMG startle response amplitudes for all 34 subjects across the 10 blocks of trials expressed in log units. Amplitude habituation across all 10 blocks was significant (F(9/297) = 14.9, p < 0.0001, E = 0.6421) with significant linear (F(1/33) = 63.8, p < 0.0001) and marginal quadratic (F(1/33) = 3.9, p < 0.06) components. The slope suggests the development of an asymptotic response level by Block 7. Fig. 2 (bottom right) shows the grand average of the EMG startle response onset latencies for all 34 subjects. The block effect across the 10 blocks (F(9/297) = 0.97) was not significant.

C. Hiram et aL/International

3.2. Vertical electro-oculogram

Journal of Psychophysiology 22 (1996) 97-109

The habituation shows a similar trend to that of the EMG and EOG startle response amplitudes (Fig. 2, bottom and middle left). Analysis of P300 amplitudes showed a significant block effect across the 10 blocks (F(9/297) = 5.8, p < 0.0001, E = 0.6642) with significant linear (F(1/33) = 20.8, p = 0.0001) and quadratic (F( l/33) = 4.8, p < 0.05) components. Fig. 2 (top right) shows the grand average of P300 peak latencies for the 34 subjects across the 10 blocks of trials. The block effect across the 10 blocks (F(9/297) = 1.2) was not significant.

(EOG) habituation

Fig. 2 (middle left) shows the grand average of EOG startle response peak amplitudes for all 34 subjects across the 10 blocks of trials. The habituation shows a similar trend to that of the EMG startle response amplitude (Fig. 2, bottom left). Analysis of the EOG amplitudes showed a significant block effect across the 10 blocks (F(9/297) = 14.7, p < 0.0001, E = 0.5803) with significant linear and quadratic components (F( l/33) = 5 1.O, p < 0.0001 and F(1/33) = 5.7, p = 0.02). Fig. 2 (middle right) shows the grand average of the EOG startle response peak latencies for all subjects across the 10 blocks. The block effect was significant across the 10 blocks (F(9/297) = 3.0, p = 0.001, E = 0.6413). There was a statistically significant cubic component (F(1/33) = 24.9, p < 0.0001) but no significant linear or quadratic component (F(l/33) = 0.45, ns and F(1/33) = 3.55, p = 0.07, respectively).

3.4. Age effects on startle and P300 amplitude latency

Fig. 2 (top left) shows the grand average of P300 peak amplitudes for all subjects across the 10 blocks. RELATED

GRAND

POTENTIALS ACROSS

AVERAGES BLOCK

(AT ~2)

AND STARTLE

THIRTY

RESPONSES

FOURNORMAL BLOCK

1

I -100

and

For the first trial-block, i.e. before habituation developed, Pearson correlation coefficients were computed between age (in months) and each of the following variables: startle amplitudes from orbicularis oculi EMG and vertical EOG, P300 peak amplitude, vertical EOG peak latency, and P300 peak latency. Only P300 peak latency showed a significant correlation with age (r = - 0.38, N = 34, p < 0.05). Within an age span ranging between 85 and 143 months (mean 116 months, standard deviation

3.3. P300 habituation

EVENT

103

BOYS

7

I

0

100

200

300

Milliseconds

400

500

-100

0

100

200

300

400

500

Milliseconds

Fig. 1. Grand averages of EEG (Pz referred to the right mastoid), vertical EGG (DC recording). and orbicularis oculi EMG in response to 104 dB noise burst across all 34 subjects for the first and seventh trial blocks. Vertical axes for EEG and EGG channels are in ~LV. Vertical axis for EMG is in log-transformed A/D units. Vertical dashed lines indicate onset of noise burst and peak latency of P300 at Pz. Horizontal dashed lines indicate baselines established 120 ms prior to onset of noise burst.

C. Hirano et al./lntermtioml

104

Journal of Psychophysiology

Latency Habituation

Amplitude Habituation

ijiL;ih

12 3 4 5 6 7 8 910

Startle (EMG)

1 2 3 I 5 8 7 8 9 10

Startle

47

(EMG

Onset)

1

Ttibmd

Tri&mck

Fig. 2. Peak amplitudes of P300, vertical EGG and orbicularis oculi EMG (left side), and peak latencies of P300 and vertical EGG and onset latency of orbicularis oculi EMG (right side) in response to startling stimuli for 10 successive blocks of four trials

each, averaged over all 34 subjects. 19.6 months), P300 peak latency shortened with increasing age. Pearson correlation coefficients were also computed between age (in months) and the linear components of habituation of the startle and P300 amplitudes and the vertical EOG and P300 peak latencies. There were no significant correlations with age within the age range of these subjects. 3.5. Relationships habituation

between startle response and P300

3.5.1. Within-subject association between the linear component of startle and P300 habituation To investigate possible relationships between the initial levels and habituation of P300 and the startle

22 (1996) 97-109

response the covariances between the habituation of the EMG or EOG startle amplitude and P300 amplitude were studied in the random regression model. For these analyses, the data were transformed to normal scores. In preliminary analyses, repeated measures ANOVAs confirmed the assumption of no significant nonlinear trends for the first consecutive seven trial-blocks for P300 amplitude and the EMG and the EOG startle amplitudes (for all nonlinear trends, F(l/33) 5 3.6, p 2 0.06). For these variables, using the first seven trialblocks, there were no significant associations between initial values or habituation of the startle response amplitudes and the initial value or habituation of P300 amplitude (2 I 1.27, p 2 0.05). Hence, initial size and habituation of P300 amplitude seemed to be independent of the initial size and habituation of the startle response. Analyses of EMG startle onset latency and EOG startle peak latency habituation across the first seven trial-blocks gave results (significant nonlinear trends and/or negative variance component estimates) that suggested that the random regression model was not suitable. For those variables, Pearson correlation coefficients were computed between EMG startle onset latency and EOG startle peak latency and P300 peak latency during the first trial-block. While there was no significant correlation (r = 0.19, N = 34, ns) between P300 peak latency and EMG startle onset latency, there was a borderline significant correlation between P300 peak latency and EOG startle peak latency (r = 0.34, N = 34, p = 0.05). 3.5.2. Evaluation of the time-course of habituation of P300 and startle: non-linear components of habituation Perusal of Fig. 2 (left side) suggests a similar pattern to the response amplitude declines for the two startle measures and the P300, with a tendency toward an asymptotic level of response at trial-block 7 and a slight increment in response amplitude at trial-block 8 for all three curves. There are also seeming differences in the three curves at trial-block 2 (an increment in P300 amplitude) and trial-block 10 (a decrement in EOG amplitude). To evaluate the validity of similarities and differences in the pattern of the three response declines, the occurrence of non-linear components received further evaluation.

C. Hirano et al./Intermtional

Journal of Psychophysiology22

As described earlier, there was a marginal quadratic component to the habituation of the orbicularis oculi EMG and a significant quadratic component to the habituation of the vertical EOG and the P300 across all 10 trial-blocks. This quadratic component was not evident across the first seven successive trial-blocks for either startle or P300 response amplitudes. The presence of a quadratic component can be used as an estimate of the development of a decreasing rate of response decline, leading to an asymptotic level of response to a repetitive stimulus. Hence, the habituation of startle and P300 response amplitudes for the first eight and the first nine successive trial-blocks was also assessed. Across both the first eight and the first nine successive trial-blocks there were strong linear components (F( l/33 2 15.0, p I 0.0005) for both the two startle measures and the P300. For the orbicularis oculi EMG, the quadratic component was marginal (F(1/33) = 3.3, p = 0.08) across nine trial-blocks and significant across eight trial-blocks (F(1/33) = 4.4, p = 0.04). For the vertical EOG, the quadratic component persisted across both nine and eight trial-blocks (F(1/33) = 6.1, p = 0.02 and F(1/33) = 4.0, p = 0.05, respectively). For P300, the quadratic component also persisted across both nine and eight trial-blocks (F(1/33) = 4.7, p < 0.05 and F(1/33) = 5.7, p = 0.02, respectively). In summary, for both the two startle measures and P300, quadratic effects of similar magnitude were present across the first 10, nine, or eight successive trialblocks, and these effects disappeared for both startle and P300 across the first seven successive trialblocks. Hence, the impression from Fig. 2 (left side) of a similar rate of response decline tending toward asymptotic levels for both startle and P300 habituation after the seventh trial-block is validated by the appearance of significant quadratic trends after trialblock 7.

4. Discussion In this group of 34 normal healthy school-aged boys, both the startle response and the accompanying P300 habituated significantly across 40 repetitive stimulations with similar response declines toward asymptotic levels developing after about 28 trials (seven trial-blocks) for both variables. These find-

105

(1996) 97-109

ings are similar to those of Putnam and Roth (1990) whose data suggest that both startle and P300 habituation reached asymptotic levels after about the same number of repetitive stimuli in adults. There was, however, no association, i.e. no covariance, within individual subjects of either the initial level or rate of habituation of the startle and the subsequent P300 response amplitudes. Thus, a particular subject could have, for example, large startle blinks and small P300 responses or slow startle habituation and rapid P300 habituation. The magnitudes or rates of habituation of one variable did not predict the other. Hence, while analyses of non-linear components of habituation across subjects revealed similar patterns of habituation for both startle and P300, analyses of linear components of habituation and initial response magnitudes within subjects revealed no association of the startle response and the subsequent P300. These results suggest both that the P300 response to a startling stimulus is not a component of the startle response (within-subject independence of linear P300 and startle parameters) and that the P300 is modulated by habituation in the same manner as the startle response itself (similar rate of response decline with asymptotic levels approached after the same number of trial-blocks).

4.1. Modulation

of P300 accompanying

startle

Sugawara et al. (1994) have previously described prestimulation-induced modulation of the automatically elicited P300 accompanying startle. The prestimulation-induced modulation of both startle and P300 included both inhibitory and facilitatory effects. Startle inhibition by prestimulation is a brainstem function involving the early low-level processing of sensory input. This type of inhibition is mediated by an inhibitory pathway in the mesopontine lateral tegmental area as demonstrated by lesion (Leitner et al., 1981; Leitner and Cohen, 1985) and stimulation (Saitoh et al., 1987) studies in the rat. This pathway, which parallels the primary startle pathway in the brain-stem (Davis et al., 1982a), impinges on the latter at, or prior to, the medial pontomedullary reticular formation (Wu et al., 1988). Hence, the mesopontine tegmental neuronal circuitry that mediates prestimulation-induced inhibition of

106

C. Hirarw et al./lntemationul

Journal of Psychophysiology 22 (1996) 97-109

startle in the brain-stem had a similar effect on the P300. The current data show that the automaticallyelicited P300 accompanying startle is also modulated by habituation. The functional neuroanatomy of startle modulation by habituation is not as specifically delineated as that underlying prestimulation-induced inhibition of startle. However, studies of the effects of human frontal lobe lesions on the habituation of awareness in peripheral vision (Troxler fading) suggest that frontal cortex increases habituation of attention to non-novel stimuli (Mennemeier et al., 1994), perhaps through selective reduction of cortical inhibition of the nucleus reticularis of the thalamus, resulting in inhibition of thalamic transmission and increased habituation (Watson et al., 1981). Combined event-related potential and magnetic resonance imaging studies during aging show that both amplitude of P300 recorded at Pz and the number of startle blinks elicited by startling stimuli during passive automatic conditions, i.e. during automatically elicited attention, correlate significantly with frontal but not parietal lobe gray matter volumes (Ford et al., 1994). Hence, our finding of a similar course of habituation for both startle and P300 (recorded at Pz> amplitudes suggests that both habituation processes are mediated by frontal cortex activity. Both the time-course of habituation of the reflex motor (the startle blink) response to and the cognitive (the P300) evaluation of the repetitive startling stimulus are modulated by the same fronto-cortical activity. In a previous study of prestimulation-induced modulation of P300 and startle, Sugawara et al. (1994) postulated a ‘bottom-up’ mode of sensory processing in which the evaluation of the startling stimulus, indexed by P300 amplitude, obeyed the rules of startle modulation by brain-stem mechanisms, i.e. prepulse inhibition. In this study, it seems that modulation of P300 and startle by habituation involves a ‘top-down’ modulation by fronto-cortical mechanisms. 4.2. Relationship startle response

of P300

to startle

stimulus

and

Prior to habituation, the magnitudes of the startle response and the P300, measured either during the first trial-block or computed as the initial level in the

computations for the random coefficient linear regression procedure, showed no significant correlations or associations. Although the latencies of startle and P300 did not meet the statistical assumptions requisite for study in the random regression model, it was feasible to compute the correlation coefficients amongst the latency values during the first trial-block, providing a measure of these relationships before habituation occurred. The P300 peak latency showed no significant correlation with startle onset latency, but did show a modest marginally significant positive correlation with the vertical EOG startle peak latency. It should be noted that the onset latency (calculated from the orbicularis oculi EMG) represents the earliest manifestation of lid movement (it may precede measurable change in the vertical EOG by 5-10 ms>. The vertical EOG peak latency, however, coincides with the termination of orbicularis oculi EMG activity (this relationship can be seen in Fig. 1) and is a measure of the time of completion of lid closure. Since the P300 peak latency is proportional to and serves as a measure of the time required for stimulus evaluation (Kutas et al., 1977; and see reviews in Donchin, 1981; Pritchard, 198 1; Howard and Polich, 19851, it seems that for this automatically-elicited P300, a part of such evaluation may include the time from the completion of the reflex response (the startle blink) to the stimulus. It is possible, then, that for sudden intense stimuli that evoke a reflex motor response, the P300 is indexing the evaluation not only of the stimulus but also of the subsequent reflex motor response to the stimulus. For strong sudden stimuli, capable of inducing large startles, the automatic P300 may reflect evaluation of, or allocation of attentional resources to, the subject’s own behavior in response to the stimulus. As the habituation process reflects the reduction of surprise induced by the startling stimulus when it occurs repetitively at a low rate of presentation in a context in which it is ignored and is not task-relevant, both the startle response itself and the subsequent mental evaluation of the importance of the startling stimulus (reflected in the P300) are reduced together to asymptotic levels at about the same time. The evaluation of the ‘importance’ of the startling stimulus may also take account of and reflect the subject’s reflex motor response to that stimulus, at least for the initial stimuli. The lack of

C. Hirano et al./Internationd

Journal of Psychophysiology 22 (1996) 97-109

significant within-subject association between the rates of habituation of startle and P300 amplitudes suggests the presence of other yet to be defined factors influencing both the individual subject’s reflex response and the subsequent cognitive evaluation of stimulus and response. Since the magnitudes and rates of habituation of startle and the subsequent P300 vary independently within the individual subject, it seems that each individual has unique startle response characteristics and that the P300 reflects an evaluation of the startling stimulus and perhaps the subsequent behavior rather than being a component of the startle response. 4.3. Intrinsic and extrinsic mechanisms of startle and P300 habituation

Startle habituation may be mediated by mechanisms that are intrinsic and/or extrinsic to the stimulus-response pathway; intrinsic mechanisms are considered part of the reflex pathway itself, while extrinsic mechanisms tend to be components of modulatory systems (Davis and File, 1984). In the rat, lesion (Leaton et al., 1985) stimulation (Davis et al., 1982b) and neuropharmacological studies (Kehne and Davis, 1984; Koch and Friauf, 1995) suggest that short-term (within session) habituation of startle is mediated by intrinsic mechanisms. Individual subject differences in mechanisms intrinsic to the startle pathway might explain the lack of association between the startle response and the accompanying P300 that may be subject to modulatory mechanisms extrinsic to the direct startle pathway. However, the similar timecourse of startle and P300 amplitude habituation, suggested by the similar times of development of quadratic trends in the habituation curves, suggests the possibility that modulatory extrinsic mechanisms may be superimposed on intrinsic mechanisms of startle habituation. 4.4. Maturational

issues

The decrease in P300 peak latency with increasing age in this school-age population is consistent with the results of studies of the P300 evoked at Pz by auditory stimuli designated as target stimuli in the conventional odd-ball paradigm across the age-range from early childhood to young adulthood (Polich et

107

al., 1990; Fuchigami et al., 1993). Additionally, Polich et al. (1990) showed that the decrease in P300 latency in response to target stimuli was correlated with the age-associated increase in memory span. Very recently, Fuchigami et al. (1995) showed a decrease in P300 latency with age in children in response to rare (non-target) tones while subjects read a book and ignored the sounds (a passive attention condition). Similarly, the current data show the same maturationally determined decrease in P300 latency during a non-cognitive, non-task automatic elicitation of P300, where the P300 is neither reflecting a cognitive decision (active attention to target) about the stimulus nor passive attention (rare vs. frequent stimuli) to the stimulus, but rather an evaluation of how surprising, i.e. how unexpected, it is. The absence of maturational change in P300 amplitude in the automatic paradigm is also consistent with maturational studies of P300 amplitude in the odd-ball paradigm (Fuchigami et al., 1993) except when target-stimuli are presented at a very low probability, in which case P300 amplitude increases with age (Polich et al., 1990).

Acknowledgements

This investigation was supported by NICHHD Grant HD-14193 and the Alice and Julius Kantor Charitable Trust.

References Achenbach, T.M. and Edelbrock, C.S. (1983) Manual for the Child Behavior Checklist and Revised Child Behavior Profile, University Associates in Psychiatry, Burlington, VT. Achenbach, T.M. and Edelbrock, C.S. (1986) Manual for the Teacher’s Report Form and Teacher Version of the Child Behavior Profile, Department of Psychiatry, University of Vermont, Burlington, VT. Bryk, A.S. and Raudenbush, SW. (1992) Hierarchial Linear Models: Application and Data Analysis Methods, Sage Publications, Newbmy Park, CA. Conover, W.J. and Iman, R.L. (1981) Rank transformation as a bridge between parametric and nonparametric statistics. Am. Statist., 35: 124- 133. Davis, M. and File, S.E. (1984) Intrinsic and extrinsic mechanisms of habituation and sensitization: Implications for the design and analysis of experiments. In: H.V.S. Peeke and L.

108

C. Hiram

et al./Intermtional

Journal

Petrinovich (Eds.), Habituation, Sensitization, and Behavior, Academic Press, New York, pp. 287-323. Davis, M. and Heninger, G.R. (1972) Comparison of response plasticity between the eyeblink and vertex potential in humans. Electroencephalogr. Clin. Neurophysiol., 33: 283-293. Davis, M., Gendelman, D.S.. Tischler, M.D. and Gendelman, P.M. (1982a) A primary acoustic startle circuit: lesion and stimulation studies. J. Neurosci., 2: 791-805. Davis, M., Parisi, T., Gendelman, D.S., Tischler, M. and Kehne, J.H. (1982b) Habituation and sensitization of startle reflexes elicited electrically from the brainstem. Science, 218: 688-690. Dixon, W.J. (1992) BMDP Statistical Software Manual, Vol. 2, University of California Press, Los Angeles, 1436 pp. Donchin, E. (198 I) Surprise! . . . Surprise? Psychophysiology, 18: 493-5 13. Donchin, E. and Coles, M.G.H. (1988) Precommentary: Is the P300 component a manifestation of context updating? Behav. Brain Sci., 11: 357-374. DSM-III-R (1987) Diagnostic and Statistical Manual of Mental Disorders-Third Edition Revised, American Psychiatric Association, Washington, DC. Ford, J.M. and Pfefferbaum, A. (1991) Event-related potentials and eyeblink responses in automatic and controlled processing: effects of age. Electroencephalogr. Clin. Neurophysiol., 78: 361-377. Ford, J.M., Sullivan, E.V., Marsh, L., White, P.M., Lim, K.O. and Pfefferbaum, A. (1994) The relationship between P300 amplitude and regional gray matter volumes depends upon the attentional system engaged. Electrocncephalogr. Clin. Neurophysiol., 90: 214-228. Fuchigami, T., Okubo, O., Fujita, Y., Okuni, M., Noguchi, Y. and Yamada, T. (1993) Auditory event-related potentials and reaction time in children: Evaluation of cognitive development. Dev. Med. Child Neural., 35: 230-237. Fuchigami, T., Okubo, O., Ejiri, K., Fuji@ Y., Kohira, R., Noguchi, Y., Fuchigami, S., Hiyoshi, K., Nishimura, A. and Harada, K. (1995) Developmental changes in P300 wave elicited during two different experimental conditions. Pediatr. Neural., 13: 25-28. Graham, F.K. (1975) The more or less startling effects of weak prestimulation. Psychophysiology, 12: 238-248. Howard, L. and Polich, J. (1985) P3C0 latency and memory span development. Dev. Psychol., 21: 283-289. Jennrich, R., Sampson, P. and Frane, J. (1985) BMDP2V. In: W.J. Dixon (Ed.), BMDP Statistical Software, University of California Press, Berkeley, CA, p. 359. Johnson, Jr., R. (1988) The amplitude of the P300 component of the event-related potential: review and synthesis. In: P.K. Ackles, J.R. Jennings and M.G.H. Coles @is.), Advances in Psychophysiology, Vol. 3, JAI Press, Greenwich, CT, pp. 69-138. Kehne, J.H. and Davis, M. (1984) Strychnine increases acoustic startle amplitude but does not alter short-term or long-term habituation. Behav. Neurosci., 6: 955-968. Koch, M. and Friauf, E. (1995) Glycine receptors in the caudal pontine reticular formation: am they important for the inhibition of the acoustic startle response? Brain Res., 671: 63-72. Kutas, M., McCarthy, G. and Donchin, E. (1977) Augmenting

ofPsychophysiology

22 (1996)

97-109

mental chronometry: The P300 as a measure of stimulus evaluation time. Science, 197: 792-795. Laird, N.M. and Wane, J.H. (1982) Random effects models for longitudinal data. Biometrics, 38: 963-974. Lammers, W.J. and Badia, P. (1989) Habituation of P300 to target stimuli. Physiol. Behav., 45: 595-601. Larson, L.-E. (1956) The relation between the startle reaction and the non-specific EEG response to sudden stimuli with a discussion on the mechanism of arousal. Elcctrocncephalogr. CIin. Neurophysiol., 8: 63 l-644. Larson, L.-E. (196Oa) Correlation between the psychological significance of stimuli and the magnitudes of the startle blink and evoked EEG potentials in man. Acta Physiol. Scand., 48: 276-294. Larson, L.-E. (196Ob) Sensitization of the startle blink and nonspecific electroencephalographic response. Electroencephalogr. Clin. Neurophysiol., 12: 727-733. Leaton, R.N., Cassella, J.V. and Borszcz, G.S. (1985) Short-term and long-term habituation of the acoustic startle response in chronic dccerebrate rats. Behav. Neurosci., 99: 901-912. Lehmamr, E. (1975) Nonparametrics: Statistical Methods Based on Ranks, Holden-Day, San Francisco. Leitner, D.S. and Cohen, M.E. (1985) Role of the inferior colliculus in the inhibition of acoustic startle in the rat. Physiol. Behav., 34: 65-70. Leitner, D.S., Powers, A.S., Stitt, C.L. and Hoffman, H.S. (1981) Midbrain reticular formation involvement in the inhibition of acoustic startle. Physiol Behav., 26: 259-268. Lew, G.S. and Polich, J. (1993) P300, habituation, and response mode. Physiol. Behav., 53: 11 I- 117. Memremeier, M.S., Chatterjee, A., Watson, R.T., Wertman, E., Carter, L.P. and Heilman, K.M. (1994) Contributions of the parietal and frontal lobes to sustained attention and habituation. Neuropsychologia, 12: 703-7 16. Gmitz, E.M., Guthrie, D., Kaplan, A.R., Lane, S.J., and Norman, R.J. (1986) Maturation of startle modulation. Psychophysiology, 23: 624-634. Plourde, G., Joffe, D., Villemure, C. and Trahan, M. (1993) The P3a wave of the auditory event-related potential reveals registration of pitch change during sufentanil anesthesia for cardiac surgery. Anesthesiology, 78: 498-509. Polich, J. (1987) Comparison of P300 from a passive tone sequence paradigm and an active discrimination task. Psychophysiology, 24: 41-46. Polich. J. (1989a) P300 from a passive auditory paradigm. Electroencephalogr. Clin. Neurophysiol., 74: 3 12-320. Polich, J. (1989b) Habituation of P300 from auditory stimuli. Psychobiology, 17: 19-28. Polich, J. (1989~) Frequency, intensity, and duration as determinants of P300 from auditory stimuli. J. CIin. Neurophysiol., 6: 277-286. Polich, J. and McIssac, H.K. (1994) Comparison of auditory P300 habituation from active and passive conditions. Int. J. Psychophysiol., 17: 25-34. Polich, J., Ladish. C. and Bums, T. (1990) Normal variation of P300 in children: age, memory span, and head size. lnt. J. Psychophysiol., 9: 237-248. Polich, J., Eischen, S.E. and Collins, G.E. (1994) P300 from a

C. Hiraw et al./lnternational

Journal of Psychophysiology 22 (1996) 97-109

single auditory stimulus. Electroencephalogr. Clin. Neurophysiol., 92: 2.53-261. Pritchard, W.S. ( 198 1) Psychophysiology of P300. Psychol. Bull., 89: 506-540. Putnam, L.E. and Roth, W.T. (1987) Heart rate responses to startling stimuli that evoke P300: Evidence for a defensive response? [Abstract]. Psychophysiology, 24: 607-608. Putnam, L.E. and Roth, W.T. (1990) Effects of stimulus repetition, duration and rise time on startle blink and automatically elicited P300. Psychophysiology, 27: 275-297. Ritter, W., Vaughan, Jr., H.G. and Costa, L.D. (1968) Orienting and habituation to auditory stimuli: .a study of short term changes in average evoked responses. Electroencephalogr. Clin. Neurophysiol., 25: 550-556. Roth, W.T., Dorato, K.H. and Kopell, B.S. (1984) Intensity and task effects on evoked physiological responses to noise bursts. Psychophysiology, 21: 466-481. Routh. D.K., Schroeder, C.S. and O’Tuama, L. (1974) Development of activity level in children. Dev. Psychol., 10: 163- 168. Saitoh, K., Tiison, H.A., Shaw, S. and Dyer, R.S. (1987) Possible role of the brainstem in the mediation of prepulse inhibition in the rat. Neurosci Lett., 75: 216-222. Sakano, N. and Pickenhain, L. (1968) Evoked response and startle blink to strong acoustic stimuli of different signal meaning. Psychophysiology, 5: 1- 14. Squires, N.K., Squires, KC., and Hillyard, S.A. (1975) Two varieties of long-latency positive waves evoked by unpredictable auditory stimuli in man. Electroencephalogr. Clin. Neurophysiol., 38: 387-401.

109

Sugawara, M., Sadeghpour M., De Traversay, J. and Chnitz, E.M. (1994) Prestimulation-induced modulation of the P300 component of event related potentials accompanying startle in children. Electroencephalogr. Clin. Neurophysiol., 90: 201-213. Ullmann, R.K., Sleator, E.K. and Sprague, R.L. (1984) A new rating scale for diagnosis and monitoring of ADD children. Psychopharmacol. Bull., 20: 160-164. Ullmann, R.K., Sleator, E.K. and Sprague, R.L. (1985) Introduction to the use of ACTeRs. Psychophannacol. Bull., 21: 915-920. Watson, R.T., Valenstein E. and Heilman K.M. (1981) Thalamic neglect: Possible role of the medial thalamus and nucleus reticularis in behavior. Arch. Neural., 38: 501-506. Wechsler, D. (1974) Wechsler Intelligence Scale for Children Revised, Psychological Corporation, New York. Welner, Z., Herjanic, B., Jung, K. and Amado H. (1987) Reliability, validity, and child agreement studies of the Diagnostic Interview for Children and Adolescents (DICA). J. Am. Acad. Child Adolesc., 44: 649-653. Wesensten, N.J., Badia, P. and Harsh, J. (1990). Time of day, repeated testing, and interblock interval effects on P300 amplitude. Physiol. Behav., 47: 653-658. Wu, M.-F., Suzuki, S.S. and Siegel, J.M. (1988) Anatomical distribution and response patterns of reticular neurons active in relation to acoustic startle. Brain Res., 457: 399-406. Yingling. C.D., Hosobuchi, Y. and Harrington, M. (1990) P300 as a predictor of recovery from coma. Lancet, 336: 873.