Quantitative sensory assessment in toxicology and occupational medicine: applications, theory, and critical appraisal

Letters, 43 (1988) 321-343

Toxicology

321

Elsevier

TXL 02048

Quantitative sensory assessment in toxicology and occupational medicine: applications, theory, and critical appraisal Jacques P .J. Maurissen The Dow Chemical Company, Midland, MI (U.S.A.) (Received

20 January,

(Revision

received

(Accepted

Toxicology Research Laboratory,

1988)

1988)

Key words: Psychophysics; Forced-choice

and Environmental

1988)

8 April,

23 June,

Mammalian

Threshold;

paradigm;

Rating

Signal detection

paradigm;

Up-down

theory;

Response

rule; Up-down

paradigm;

transformed

Yes-no

paradigm;

rule; PEST

SUMMARY systems

are affected

with psychophysical

Sensory

methods.

physical

dimension(s)

of a stimulus

and the behavioral

tion of psychophysics

to toxicology

is illustrated

vision, and somatosensory and followed by delivering the subject. method

sensitivity.

quantified

stimuli

Psychophysical

Sensory

specify

of constant

The three major

processes

study

response

can be quantitatively

of the relationship

it generates.

with a few examples foundations.

in a prespecified

taken

order and by requesting

response

can be studied

evaluation of stimuli,

terminology

illustrate

the potential

are the yes-no,

as well as in animals

and methodology.

Finally,

brought

a few suggestions

about

from the

methods.

The

the subject has to make to stimuli presented

in humans

confusion

response and include

and tracking forced-choice,

accord-

and rating

by several techniques,

as classical conditioning, operant behavior, and reflex modulation. A critical review of the application of psychophysical methods and response Several examples

is presented

is accomplished

a standardized

of presentation

paradigms

the

contribu-

from the fields of audition, of psychophysics

as well as the adjustment

response

assessed

between

First, the unique

Psychophysical

the sequence

stimuli,

specify the standardized

format.

Their functions

is the scientific

Then, a brief survey of the evolution

methods

the method

paradigms

ing to a specific paradigms.

of chemicals.

Psychophysics

by a review of its basic scientific

of limits,

response

by a number

paradigms

by unorthodox

such

is developed.

use of psychophysical

are given for the evaluation

of sensory

pro-

cesses in humans.

Address Toxicology,

for correspondence: J.P.J. Maurissen, Midland, MI 48674, U.S.A.

Abbreviations:

UDTR,

up-down

transformed

Ph.D., rule; PEST,

The Dow Chemical parameter

estimation

Company,

1803 Bldg -

by sequential

testing.

322

INTRODUCTION

Sensory systems provide the interface between the living organism and the external milieu, and represent targets of pathological processes affecting the nervous system in animals and humans. Endogenous diseases as well as exogenous events (physical or chemical) can alter the structure and function of sensory receptors, afferent pathways, and primary projection areas in the brain. It is important to evaluate the integrity of sensory systems from a functional point of view. Solely relying on a morphological approach presents some inherent problems. A morphopathological evaluation is destructive and does not usually lend itself to repeated measures in the same organism. Sophisticated morphological techniques (e.g., electron microscopy) can also generate data difficult to interpret. Perturbation of sensory functions may not be detected by simple clinical observation. More elaborate techniques are necessary. In this context, psychophysics provides exquisite and robust tools to quantify sensory functions in animals and humans. The intent of this paper is 4-fold. (1) Illustrate the contribution of psychophysical (i.e., sensory) studies to toxicology with a few examples. (2) Recount briefly the salient events forging the history of psychophysics, and to define its scope. (3) Acquaint the reader with the (often misunderstood) methods, paradigms, and techniques for the assessment of sensory functions in humans and animals. (4) Evaluate critically the methodology used in human sensory studies as applied to toxicology

and occupational

NEUROTOXICANTS

A large number

AND SENSORY

of chemicals

medicine. FUNCTIONS:

affect

SELECTED

sensory

organs

EXAMPLES

and functions

[l-3].

Some

overviews have focused more specifically on the psychophysical approach to chemically induced auditory [4], visual [S], somatosensory [6,7], and olfactory disorders [8]. The intent of this section is not to review this literature exhaustively, but to show some examples of the unique contributions of psychophysics to toxicology. Audition The ototoxic effects of toluene were unknown until Pryor et al. described them in 1983 191. The authors exposed rats to 1400 and 1200 ppm of toluene, 14 h/day, 7 days/week for 5 weeks. Rats were trained to avoid an electrical shock delivered to the floor by climbing a pole suspended from the ceiling of the cage. The aversive stimulus was preceded by a tone. Climbing the pole during this warning tone ter-

323

minated the trial and reflected detection of the auditory stimulus. Both intensity and frequency of the tone were varied. Comparison of the conditioned avoidance performance of the control and exposed groups revealed that toluene impaired hearing markedly at 12, 16, and 20 kHz, and slightly at 8 kHz. No difference in hearing was found between control and exposed rats at 4 kHz. These data showed for the first time that toluene preferentially affected hearing at higher frequencies in the rat, and left the low frequencies practically untouched. Subsequently, electrophysiological techniques revealed that the latency of the 8th nerve component of the brainstem auditory-evoked response was prolonged in toluene-exposed rats at high frequencies in comparison to control animals [lo]. Vision The effects of acrylamide on visual functions were first described by Merigan et al. [ 111. Macaque monkeys faced two oscilloscopes with evenly illuminated displays. One of two types of stimuli appeared randomly on either scope, while the other scope was blank; a fine grating or a flickering unpatterned stimulus was used to measure visual acuity or flicker fusion, respectively. In both cases, the monkeys received some fruit juice after pressing the button corresponding to the scope presenting the stimulus. Four monkeys were trained in both visual modalities. Three of them were administered oral doses of acrylamide (10 mg/kg/day, 5 days a week) until they showed overt signs of intoxication. Each monkey was used as its own control, and one served as a time-matched control. By the end of dosing (7-9 weeks after the first dose), visual acuity and flicker fusion were reduced in dosed monkeys. After termination of dosing, flicker fusion recovered quickly and completely. Visual acuity substantially improved, but had not completely recovered by 19-20 weeks after the end of dosing. Morphological examination of a monkey killed at the end of the dosing period indicated that the described loss of function was accompanied by distal axonal swelling in the optic tract and lateral geniculate nucleus [ 12,131. Somatosensory

sensitivity

Acrylamide causes central-peripheral distal axonopathy. Since sensory signs and symptoms usually dominate the clinical manifestations of intoxication with this chemical, Maurissen et al. [14] developed a system to assess quantitatively the effects of acrylamide on vibration and electrical sensitivity in monkeys, and to follow the functional recovery after cessation of intoxication. Six monkeys were trained to report detection of a vibratory or electrical stimulus applied to the fingertip. After baseline data were obtained, four monkeys were dosed orally with 10 mg/kg of acrylamide 5 days/week until the appearance of overt

324

toxic signs. Each monkey served as its own control. The other two monkeys were used as time-matched controls. Fine visuomotor coordination and body weights diminished during dosing, but had recovered within 5-8 weeks. Vibration sensitivity also decreased during the same period, but remained impaired much longer, outlasting all the other effects. Electrical sensitivity had been unaffected during the whole study. These data indicate that vibration sensitivity can trace the time course of acrylamide intoxication, and provide unique information on the clinical status of the peripheral nervous system that could not be inferred from the other observations made here. Morphological analysis of sural nerve biopsies in two exposed monkeys revealed that the extent of the damage considerably differed between the two animals (25% of nerve fibers vs. an occasional nerve fiber affected), while their sensitivity to vibration was reduced in a comparable manner. These data emphasize the importance and complementarity of a functional approach to sensory analysis. In this study, a morphological approach alone could have given a distorted conception of the clinical status of the subject. BRIEF HISTORICAL

SURVEY

The study of sense organs evolved from developments occurring mainly in the field of physiology during the 19th century. In the early 1800s special attention began to focus on the peripheral nervous system and the sense organs. Theodore Schwann, from the University of Liege, discovered the myelin sheath [1.5]. Bell [16] demonstrated that the anterior root of the spinal cord was motor in function, while Magendie [17] described the sensory function of the posterior roots. The formulation of the doctrine of specific energies, which states that a given nerve generates only one type of sensation, gave an impetus to sensory physiology during the 19th century [18]. Ernst Heinrich Weber (17951878), professor of anatomy and physiology in Leipzig, began the study of the innervation of the skin and special sense organs in humans, and investigated two-point discrimination, as well as perception of weight and temperature [19,20]. But it was Gustav Theodor Fechner (1801-1887) who really formalized the study of sense organs in his ‘Elemente der Psychophysik’ in 1860 [21]. In this book, he also described psychophysical methods still in use today. More recently, signal detection theory elaborated a theoretical framework that directly benefited psychophysics. It was developed in the 1950s by engineers who drew upon statistical decision theory and electronic communications. Signal detection theory provides a means of analyzing detection problems. It has broadened the scope of the psychophysical approach [22], and its major contribution has been to isolate the system’s intrinsic discriminative capabilities from factors affecting the translation of these discriminations into decisions. In other words, it has allowed the experimenter to tease apart true sensory events from nonsensory events that can

325

significantly contaminate the results of sensory testing. Signal detection theory has distinguished between three main experimental situations in which detection tasks can be performed (yes-no, forced-choice, and rating paradigms). These paradigms will be described later. OBJECT

OF PSYCHOPHYSICS:

MEASURING

THE

‘UNMEASURABLE’

Psychophysics is the scientific study of the relationship between a physical dimension of a stimulus and the behavioral response it generates. To describe the psychological aspect of the word ‘psychophysics’, the term ‘behavioral response’ is used here instead of ‘sensation’ to avoid introspectionistic and mental dimensions, which hamper the scientific approach. Operational definitions are indispensable to scientific endeavor. Stevens [23] has best captured the true nature of the psychophysical approach: ‘What we can get at in the study of living things

are the responses

of organisms,

physical

eludes

test.

Consequently,

- about

differential

mental

stuff,

which,

statements

about

sensation

organisms.

(...) I know nothing

by definition,

become

statements

about

about your sensations

objective responses

except what your behavior

not some hyperverifiable reactions

of

tells me.’ (p. 386)

The realm of psychophysics is still commonly misunderstood nowadays, and a recurrent fallacy is that ‘sensations’ can only be studied in humans due to their intimate nature, a mistake made by Claude Bernard himself [24]. It is, however, more surprising to read a similar statement in a more recent publication [25]: ‘Resort to animal the organoleptic

experimentation or sensory

instead

response...’

of man, eliminates

one whole area of response,

namely,

(pp. 18-19)

Basically, psychophysical evaluation is accomplished by delivering stimuli in a prespecified order (psychophysical methods) and by requesting from the subject (human or nonhuman) a codified response to the stimuli presented according to a specific format (response paradigm). PSYCHOPHYSICAL

METHODS

Psychophysical methods, established by Fechner [21], describe presentation of stimuli. Associated with each stimulus is a response function that relates these two parameters is usually characterized called the psychometric function. More details on psychophysical found in a number of textbooks [26-301.

the sequence of probability. The as ogival and is methods can be

Method of limits The method of limits consists of an equal number of presentations of ascending and descending stimulus levels. In the descending series, the first stimulus magnitude exceeds the detection level and is decreased until the first report of ‘no

326

*et

2nd

4th

set

3rd set

+

+

+

+

+

+ + +

+

+

T3 TRIALS Fig. 1. Stimulus

sequence

in the method

sequence

‘-I’

TRIALS of limits. The plus sign means ‘correct

sign means Fig. 2. Stimulus

:

‘1’:‘6

in the method

‘incorrect

of constant

the minus

stimuli.

sign means

response’,

and the minua

response’. The plus sign means ‘correct

‘incorrect

response’,

and

response’.

detection’ by the observer. In an ascending series, the first stimulus magnitude is situated below the detection threshold and is increased until the first detection response is given (Fig. 1). In each series, the midpoint between the last two levels is calculated and the threshold consists of the average of these midpoints. Two types of errors have been reported with this method in humans. One is the error of habituation, in which the observer has a tendency to repeat the preceding response, and the other is the error of anticipation, in which the observer prematurely changes the response from ‘detection’ to ‘no detection’ (or vice versa) in view of the number of stimuli already presented in this series. Starting each series at a different stimulus level is therefore mandatory. B-

16 12 z + Cl ++ + + z 8

6- ++

+ -++

++ s -

:: z z

.+-

-+++

15

10

5

++ ++

_

L

++

z -

_

0

20

sequence intensity

Fig. 4. Stimulus

sequence

the intensity

in an up-down

down

-

+.

!,-,,,

5

10

15

20

TRIALS

in the simple up-down at the next trial,

+-

2

TRIALS Fig. 3. Stimulus

-

W

1

0

++ ++

++ 4

rule: one correct

while one incorrect transformed

at the next trial,

response

response

rule: two correct

while one incorrect

(-)

(+)

responses

response

decreases

increases

(-)

the stimulus

it.

in a row (+ +) drive drives it up.

327

Method of constant stimuli This method obviates some of the errors potentially associated with the method of limits. Stimuli are delivered in a random order to prevent the observer from anticipating the magnitude of the following stimulus. Every stimulus of each set is presented for the same number of times (Fig. 2). The percentage of correct responses to each stimulus level is calculated, a psychometric function is derived, and the level corresponding to some predetermined correct response rate represents the threshold. This method, however, has an important disadvantage: every stimulus level must be tested many times before a threshold can be determined with a reasonable level of accuracy.

Method of adjustment The method of adjustment or of average error is a variant of the method of limits. The observer, most often an active participant, has direct control over the attribute of the stimulus under evaluation [31], and is requested to adjust the stimulus level to reach some criterion (e.g., to make the stimulus just detectable). One of the advantages of this method is that it provides a rapid way of obtaining thresholds. This is especially interesting when one is measuring a quickly changing threshold. However, test results can easily be biased by the subject.

Tracking methods Simple up-down

rule

The simple up-down rule is a variation of the method of limits. The order of stimulus presentation depends directly on the subject’s performance. If the observer detects the presence of the stimulus, its intensity is decreased at the next trial. If the observer does not detect the stimulus, its intensity is increased at the next trial [32]. Going

from a ‘detection’

to a ‘no detection’

(or vice versa) defines

3). With this up-down rule, all the observations on the stimulus level that is detected on 50% estimate of this stimulus level and is calculated sities [33], or the stimulus intensities associated

a reversal

(Fig.

are hovering about and converging of the trials. The threshold is an by averaging all the stimulus intenwith each reversal [34]. This simple

up-down rule has the advantage of maximizing the presentation of stimuli in the neighborhood of the threshold, and typically requires less time than the method of limits or of constant stimuli.

Up-down

transformed

rule

The up-down transformed rule (UDTR) is a generalization of the simple up-down rule, and it aggregates responses around a stimulus level correctly detected in a prespecified percentage of the trials, usually other than 50%. Figure 4 illustrates

328

:

“Q

qa5 Ql

,

-

z”’ 4

+ I

+ I +++

%

” Q”C

+‘r ’ ++

+

329

an example of a transformed rule: two consecutive correct responses will decrease the stimulus level at the next trial, while only one incorrect response is needed to increase the stimulus level at the next trial. Table I (panel A) shows that, with this rule, the series of responses will converge from any start point to a stimulus level correctly detected in 70.7% of the cases. The average stimulus level at steady state represents the threshold. Panels B and C (Table I) illustrate two additional strategies. More details on these procedures can be obtained from Wetherill and Levitt [34]. Parameter estimation by sequential testing The parameter estimation by sequential testing (PEST) method is also a variant of the up-down rule. It automatically determines each stimulus level to obtain information about the location of the threshold as efficiently as possible. Stimulus step size is variable and the testing session ends when a selected criterion has been met, that is, when some predetermined minimum step size has been reached. By the end of the session, the threshold is represented by the last stimulus level tested. Such a technique uses precise rules to determine step size and number of trials needed to calculate a threshold. This method can reduce the total number of trials to achieve some predetermined criterional stability. PEST was developed by Taylor and Creelman [35] and has been revisited on several occasions since [36-411. RESPONSE

PARADIGMS

While psychophysical methods only specify the rules governing a change in stimulus level, response paradigms define the type of response requested from the observer to some signals presented according to a specified format (i.e., one or several stimuli presented in each trial). The three major paradigms will be briefly mentioned here. Yes-no paradigm In the yes-no paradigm, the subject has to decide at each presentation whether an observation interval contains a signal (i.e., stimulus) or not (‘yes’ or ‘no’). No other response is accepted in the simple detection task. To minimize bias, the number of trials without a signal (also called ‘catch’ or ‘blank’ trials) should approximate the number of trials with a signal (‘test’ or ‘stimulus’ trials). It is advisable to let the subject know whether the response was correct or not. From these two stimulus and response alternatives (presence/absence vs. ‘yes’/‘no’), one can build a stimulus-response matrix, in which entries in each cell contain conditional probabilities (Fig. 5). With the yes-no paradigm, the criterion used by the subject to report the presence or the absence of the stimulus can vary widely among observers and even within

330

STIMULUS

F--e!-

ws %

: fn

l-----i

Fig. 5. Stimulus-response

matrix

false

hit

positive

P(Nls) miss

denoted

p(Sln)

P(SlS)

in a yes-no

P(Nln)

correct rejectton

paradigm.

Actual

by ‘s’ or ‘n’, while ‘S’ and ‘N’ refer to the reported

presence

presence

or absence

or absence

of the signal

of a signal

is

by the

observer.

observers over time. Several factors will affect this criterion. Two of these are the probability of signal presentation (i.e., the proportion of trials with and without a signal), and the payoff matrix, which assigns rewards and penalties associated with right and wrong decisions. These two nonsensory factors are under the direct control of the experimenter and directly manipulate the observer’s response. A high probability of stimulus presentation will favor ‘yes’ responses [42], and will result in an increased proportion of false positives (‘yes’ responses in the absence of a stimulus). As a consequence, it will artificially drive the ‘threshold’ to lower values if no correction is applied. Table II illustrates how false positives can affect sensory thresholds. The raw percentage of ‘yes’ responses is corrected for three falsepositive rates, using a correction described by Blackwell [43]. The thresholds are calculated here by linear interpolation between two points (for the sake of simplicity) and set as the stimulus level corresponding native ways of dealing with false positives detection theory (Ref. 22, pp. 86-116).

to the 50% detection point. Alterhave been proposed by the signal

Forced-choice paradigm In the forced-choice paradigm, the subject is presented with several intervals or alternatives, and has to indicate which one contains the stimulus. If no signal is clearly perceived, guessing is required. Abstaining from making a choice is not allowed. It is recommended that feedback be given to the subject [44]. When two alternatives are used, they may be distributed in the spatial or temporal domains and are referred to as (spatial) two-alternative forced-choice and two-interval forcedchoice paradigms, respectively. In a two-interval forced choice, for example, the signal is presented during the first or the second interval with the same probability. With this paradigm, the point

331

TABLE

II

CALCULATION

OF THE

OF FALSE-POSITIVE

50% THRESHOLD

IN THE YES-NO

PROCEDURE

AS A FUNCTION

RATES of ‘yes’ responses

corrected

for

Raw percentage

false positives

level

‘yes’ responses (hits)

10% FP

20% FP

30% FP

6

100

100

100

100

5

70

67

63

57

4

60

56

50

43

3 2

40

33

25

14

10

0

1

0 Intensity

of

Percentage

Stimulus intensity

level corresponding

3.5 a Correction

according

to 50% of ‘yes’ responses 3.7

to Blackwell

(FP)a

4.0

4.5

[43]; (p. 309).

of no detection will necessarily correspond to 50% of correct answers. The threshold is defined as the stimulus level corresponding to some response probability or discriminability, usually to the midpoint of the psychometric function, that is, to the 75% correct response rate for the two-alternative and two-interval forced-choice paradigms [35,38,40,45-551.

Rating paradigm This paradigm is a variation of the yes-no paradigm, and is designed for humans. One signal per trial is presented to the observer who is instructed to report, on a several-point scale, the likelihood that the signal or no signal was presented. TECHNIQUES

FOR ANIMAL

STUDIES

The basic psychophysical methods and response paradigms developed for a human observer are germane to those designed for lower order animals. Sensory capacities of nonverbal subjects can be assessed by several techniques: classical conditioning, operant behavior, and reflex modulation.

Classical conditioning In 1927, Pavlov published a new method for the investigation of reflexes [56]. By pairing a stimulus that had no observable effects (neutral stimulus), with another stimulus capable of eliciting a response or reflex (unconditioned stimulus), Pavlov was able to induce the same response by presenting the once neutral (now condition-

332

ed) stimulus alone. This method of pairing is called classical or Pavlovian tioning. This principle underlies the ‘method of contrast’ also developed by Pavlov

condito dif-

ferentiate a conditioned stimulus, which was always accompanied by an unconditioned stimulus and a response, from a neighboring neutral stimulus never followed by the unconditional stimulus or response. Differential responding entailed sensory differentiation. An example of the application of classical conditioning in psychophysics can be seen in Northmore and Muntz [57], who measured the sensitivity of fish to moving and stationary bars of light and to diffuse light. An electrical shock served as the unconditional stimulus. Its effect was to produce a consistent increase in the activity of the fish. By pairing neutral and unconditional stimuli a certain number of times, the neutral (now conditioned) stimulus alone produced the increased swimming activity and was evidence of detection. In a similar manner, Otis et al. [58] induced decreased respiration and heart rates by electrical stimulation in goldfish to assess detection of a change in illumination. Passe [59] used a classical conditioning technique, called autoshaping, to evaluate the absolute visual sensitivity in pigeons. In this experiment, the pigeons were fooddeprived to reach 80% of their free-feeding body weights. A response key was illuminated and followed by grain delivery. After an adequate number of trials, the pigeons reliably pecked the key only when it was illuminated, even though pecking the key was not required for grain delivery. The intensity of the key illumination was varied, and pecking the key when it was illuminated indicated detection of the light. Hearing also can be studied with classical conditioning techniques. Jamison [60] used ammonia as the unconditional stimulus to decrease heart rate in rats. After pairing a tone with ammonia presentation, the tone alone reduced heart rate. Detection of the stimulus was evidenced by a change in heart rate. Overall, there are limitations to the classical conditioning approach: total number of trials per day, necessity for reconditioning due to loss of the conditioned reflex, and problem of defining a criterional response change, to name a few.

Operant behavior In 1938, Skinner elaborated on Thorndike’s ‘law of effect’, systematized an approach to learned motivated behavior, and developed the methods of experimental analysis of behavior [61]. He explained behavior by its consequences: previous consequences model current performance. Because an operant (i.e., any response that modifies the environment) can be placed under the differential control of a stimulus, new possibilities for sensory research emerged. Several procedures have been designed to analyze sensory processes in animals, and they use one or more response manipulanda. The major drawback of this method lies in the training required to place a behavior under stimulus control.

333

Go/no-go procedure This procedure

is the operant

version

of the yes-no

paradigm.

The animal

is train-

ed to give a response in the presence of a standard stimulus and to withhold it in its absence or in the presence of a comparison stimulus. Terman and Terman [62] used this procedure to evaluate rats’ discriminative capabilities for auditory intensity. By pressing a bar, the rats turned on a tone. After the tone, a 3-s choice interval was presented. Rats received a reward (or reinforcement) if: (1) they pressed the bar during the choice interval following presentation of the tone at standard intensity, (2) they did not press the bar during the choice interval following presentation of a tone at a different intensity. Aversive stimuli can also control such a behavior. Clack and Herman [63] tested monkeys’ auditory thresholds by shock avoidance conditioning and generated data comparable to those obtained with positive reinforcement.

Forced-choice paradigm The forced-choice paradigm also has been used successfully in animal sensory research. Typically, two stimuli are presented. They can be spatially or temporally spaced, and each stimulus is associated with one response manipulandum or one interval. For example, Yager and Thorpe [48] tested color vision in goldfish in the following manner: the fish first swam to the back of the tank, made an observing response by pressing a key that illuminated both choice keys, swam toward the choice keys, and struck one of them. On the average, every third correct response was reinforced. A correction procedure prevented the formation of a position habit.

Conditioned

suppression procedure

In 1941, Estes and Skinner when a signal that terminates

described the ‘conditioned emotional response’ [64]: with a shock is superposed over a steady rate of an

operant, it acquires suppressing properties. The suppression is measured by the ratio of the response rates before vs. during the signal and reflects detection of the stimulus. However, it was not until later that this procedure was exploited in psychophysics [65,66]. Using this procedure, Sidman et al. [65] studied hearing and vision in animals. Mice were trained to lick a tube and received a drop of milk after a variable number of licks. This schedule of reinforcement generates a sustained rate of licking sufficiently stable so that the interfering effects of random signals announcing an electrical shock can be evaluated. These warning stimuli have a fixed duration and develop conditioned suppression (i.e., the rate of licking is drastically reduced during the warning stimuli and therefore indicates detection). Excellent reviews of animal psychophysics can be found in Blough and Blough [67], Stebbins [68], and Blough and Young [69].

334

RefIex modulation Hoffman phenomenon presentation,

and Ison [70] defined reflex modulation or reflex modification as ‘. ..the whereby the reflex elicited by one stimulus is modified by the prior withdrawal, or change of another (usually weaker) stimulus’.

As early as 1863, Sechenov showed that stimulation of the frog’s brain inhibited reflexes to painful cutaneous stimuli [71]. One of the first applications of reflex modification to sensory analysis goes back to Yerkes [72]. He established that ringing a bell before tapping on the frog’s head modified the flexor jerk reflex, and concluded that frogs can hear. More recently, the technique has been refined and applied by several laboratories to assess sensory processes. For example, Young and Fechter [73] investigated animal audiometry and used reflex modulation with the method of constant stimuli. They tested rats and guinea pigs with 115 dBA white noise startle bursts of 20-ms duration. On control trials, the startle-eliciting stimulus was presented alone. On prestimulus trials, a pure-tone stimulus of varying intensity and frequency preceded the startle-eliciting stimulus by 100 ms. For each test session, an average startle amplitude was calculated at each prestimulus intensity and frequency and expressed as a percentage of average startle amplitude during control trials. A psychometric function relating relative startle amplitude to prestimulus intensity was generated for each tested frequency. The data calculated for rats and guinea pigs with reflex modulation are comparable to those collected with operant techniques on rats by Kelly and Masterton [74] and on guinea pigs by Prosen et al. [75]. One of the problems of reflex modulation consists in the definition of a response change that can evidence sensory detection. More details on this technique can be found in Hoffman and Ison [70], Ison and Hoffman [76], Hoffman [77], and Fechter CRITICAL

and Young

[78].

EVALUATION

OF HUMAN PSYCHOPHYSICAL

STUDIES

Psychophysics grew roots in the middle of the 19th century, and its methods have been refined, analyzed and criticized since. Overlooking more than 100 years of scientific advancement in psychophysics, as well as neglecting the framework of signal detection theory, has fostered confusion in the application of psychophysical methods in human toxicology and occupational medicine. In the next few paragraphs, this point will be illustrated.

Nonclassical terminology It has been implied that the ‘method of constant stimuli’ is always present, when, in fact, it signifies that discrete, generated randomly, as described earlier:

means that the stimulus prespecified stimuli are

335

‘Some methods the stimulus

use constant

is always

The forced-choice ‘In the forced

paradigm

choice method

does not necessarily defined This

stimuli

present...’

stimulus

presentation.

also has been erroneously

defined

the examinee

of discrete

knows when the stimulus

know it was. In this procedure

by a light clue, the observer

(1791, P. 8) is, in fact,

instead

the accepted

In these procedures

([79], p. 9)

is required

definition

during

to indicate

of the yes-no

as follows:

may have been presented,

a fixed time interval,

whether

or not the stimulus

paradigm,

but

which is usually was present.’

which is diametrical-

ly opposite to the forced-choice paradigm. Similarly, giving a choice between two intervals and also allowing the observer to abstain from making a choice has been erroneously referred to as a forced-choice paradigm (in a commercially available operating manual). By definition, abstention is not allowed in a forced-choice situation. The ‘method of limits’ has been contrasted with the ‘forced-choice paradigm’ although these two pertain to two entirely different domains, as explained above. The former refers to a method of stimulus presentation and the latter to a response paradigm. The method of limits conceivably can be used in a forced-choice situation. What some authors really meant by ‘method of limits’ was the ‘yes-no paradigm’ used in conjunction with the method of limits. Such confusion also has been perpetuated in some brochures and operating manuals for commercially available devices.

Nonconventional

procedures

Some authors do not describe the psychophysical method or the response paradigm used in their studies [80]. Sometimes modified or new methods are adopted for which no or almost no literature exists, or for which no clear rationale is given. For example, Arezzo et al. [81] used the following up-down transformed rule to test sensation: if stimuli are correctly identified on two out of three trials, the stimulus level is decreased; if errors are made on two out of three trials, the stimulus level is increased. Binomial analysis of this strategy and of the simple up-down procedure (Table III, panels A and B) shows that both rules converge on the exact same stimulus level (i.e., the level corresponding to the 50% correct response rate). Therefore, both procedures give the same estimate, but, by definition, it will take between 2 and 3 times longer to get the same answer with the transformed strategy described above (Table III, panel B), compared to the simple up-down rule (Table III, panel A). Advantages of this up-down transformed procedure are elusive.

Unorthodox Sosenko (described

threshold definition et al. [82] and Arezzo et al. [81] used the modified up-down rule in Table III, panel B) in conjunction with the two-alternative forced-

____

_ +-T +--r PC1-PY

P(1 -PY

(1-PI2

.__-..

2PU -HZ+

----____.

6p*--4p3-l=O

..

2&l -p)+p2=

p=i-p

-

._-.-.-_ ___ Condition of convergenceb

(1 -Pf2

-_.-._____

p=o.s

p=o.s

--

Probability of correct response convergence __.-_-

‘The series converges or stabilizes at the level where the sum of the probabilities of a decrease equals the sum of the probabilities of an increase.

“p= probability of a correct response.

Two out of three incorrect responses + increased level

t

&l-P)

-++i

--

P2

P%-PI

+-+1

1-P

P

Probability of changing level”

++l

-1;

Incorrect response + increased level

B: Two out of three correct responses --*decreased level

t-l

Symbolic representation

+ decreased Level

A: Correct response

-~

UP-DOWN RULES -_-_-___ .~ Up-down rules for stimulus sequence

TABLE III

337

STIMULUS

Fig. 6. Psychometric

function

relating

detection

INTENSITY

probability

forced-choice

to stimulus

intensity

in a two-alternative

paradigm.

choice paradigm. They calculated the sensitivity threshold as the stimulus level corresponding to a 50% correct response rate (i.e., chance detection). Such a threshold definition is unusual in a two-alternative forced choice because it corresponds to a stimulus level that cannot be detected by definition. There is no single point on the abscissa of the psychometric function that corresponds to the 50% correct response rate but rather an infinity of points encompassing all potential levels that the subject could not detect (Fig. 6). This criterion would be equivalent to estimating the stimulus level corresponding to 0% detection in a yes-no paradigm. A similar approach has already been criticized by Rose et al. [47] who stated: ‘...it makes point,

no sense to use the yes-no

but a continuum

of stimulus

rule in a forced-choice

situation

since there is not just one

values for which the forced-choice

psychometric

function

has

the value of 0.5.’ (p. 200)

by Kershaw [83] who wrote: ‘In a few papers interpretation

other UDTR

is used in 2IFC experiments, or no chance

by Penner paradigm: ‘Obviously,

of detecting

estimates

the stimulus.’

it is necessary

terest as a dependent these thresholds

to track

many stimulus variable.

are not unique.

probability

associated

(...) It is difficult

that they obtain.

then levels will drift toward

low stimulus

rule

levels for which there is little

stimulus

statement

about

levels for which a unique

detection

levels result in 100% and 50% detection, We cannot

compare

thresholds

We must restrict our measures with them.’

the forced-choice

obtained to stimulus

for an invisible

probability

exists.

these levels are of no inin various

conditions

if

levels that have a unique

(p. 115)

and by Levine and Shefner [52] who wrote about the forced-choice ‘As 50% is the score expected

to see what

If the Up-and-Down

(p. 36)

[51] who made the following

Since infinitely

detection

rules are used with the 21FC procedure.

can be made of the threshold

stimulus,

it cannot

paradigm:

be taken as the threshold.’

(p. 17)

Limited number of trials Care should be taken to use an adequate number of trials to ensure threshold reliability and validity. On the basis of statistical analysis, Dixon and Mood [84]

338

recommended a sample size of 40-50 trials for the up-down method. Shelton et al. [38] showed that neither PEST nor the transformed up-down rule will converge on a threshold with only ten trials and recommend threshold with the up-down transformed rule. their tests after a total of five incorrect average Some earlier, absence

responses

running 40-60 trials to evaluate a However, some authors conclude and report

their thresholds

as the

of eight values [82]. authors have used only 20% of catch trials per session [85]. As discussed such a practice will most likely underestimate the actual threshold in the of correction.

Nontraditional

units

Physical parameters of stimuli should be fully characterized and calibrated so that similar conditions can be replicated in a different setting. Vibration sensitivity amplitude thresholds have been expressed in inappropriate and/or erroneous units, such as ‘Biothesiometer units’, or more simply, ‘units’ [82,86], ‘volts’ [8.5,87-931, ‘volts square’ [94] and even in ‘Hertz’ [95]. In the field of vision, it is not unusual to see light flash intensity characterized only in terms of a device setting on a specified photostimulator rather than in appropriate optical units 1961. The experimenter should measure the stimulus of interest rather than an attribute generated before or during the production of this stimulus. For example, Bleecker [88] and Anger et al. [91] studied vibration sensitivity, but they measured the voltage applied to the vibrator rather than the actual vibration amplitude. With the same device, the former reported a threshold of 3.67 volts in normal subjects of 31-40 years of age, while the latter recorded 8.8 volts in a reference group with an average age of 36 years. As a test score of about 7 volts should be considered as significantly elevated according to the nomograms furnished by the manufacturer, a value of 8.8 volts would certainly have to be considered as highly pathological although it characterizes a control group. More likely, this difference may be related to lack of calibration of the two units. CONCLUSION

AND RECOMMENDATIONS

Psychophysical studies have helped characterize the toxic potential of chemicals to sensory functions in a quantitative and noninvasive manner. Extrapolation of psychophysical observations from one species to another is facilitated because sensory pathways are phylogenetically old structures, and very similar methods can be used across species. This psychophysical approach has appealed to many, and a large number of studies have flourished. On several occasions, however, sensory testing in humans has presented deficiencies not usually present in most animal studies. Therefore, it is suggested that major methodological features developed for sensory assessment

339

in animals (including signal detection theory approach) be taken into consideration in human sensory analysis. The following recommendations should be considered. (1) Minimize evaluation. (2) Automate

relations

between

stimulus

the

delivery

tester

and

and response

the

testee

during

sensory

recording.

(3) Standardize instructions. (4) Preestablish rigid objective rules to define all the test events (i.e., response, stimulus sequence,. . .), and to accept or reject data points. (5) Choose appropriate units of measurement that will allow for comparison between different studies. (6) Describe all relevant physical dimensions of stimuli, and perform frequent calibrations. (7) Insert catch trials (whenever appropriate) in adequate proportions to correct sensory data. (8) Use appropriate number of trials. (9) Give feedback on subject’s performance. (10) Use correction procedures to prevent position habits, whenever applicable. All these factors will minimize errors and potential unconscious bias on the part of the subject and/or test administrator, and will increase the reliability, accuracy, and validity of the data collected.

REFERENCES

1 Merigan, York,

W.H.

and Weiss, B. (Eds.) (1980) Neurotoxicity

of the Visual System,

Raven

Press,

New

Special Senses, Raven Press,

New

NY.

2 Hayes,

A.W.

York,

NY.

3 Miller,

J.J.

4 Stebbins,

(Ed.) (1985) Toxicology (1985) Handbook

W.C.

behavioral

(1982)

toxicology.

5 Merigan,

W.H.

of the Eye, Ear, and Other

of Ototoxicity,

Concerning Environ.

(1979) Effects

CRC Press,

the need

Health

for

Perspect.

of toxicants

more

Boca Raton, sophisticated

FL. animal

models

in sensory

44, 77-85.

on visual

systems.

Neurobehav.

Toxicol.

1 (Suppl.

l),

15-22. 6 Maurissen, (Suppl.

J.P.J.

7 Maurissen,

J.P.J.

Neurobehav. 8 Halpern,

of toxicants

(1985) Psychophysical

Toxicol.

Teratol.

on the somatosensory

system.

Neurobehav.

Toxicol.

1

testing

in human

populations

exposed

to neurotoxicants.

7, 309-317.

B.P. (1982) Environmental

Perspect. 9 Pryor,

(1979) Effects

l), 23-31.

factors

affecting

chemoreceptors:

an overview.

Environ.

Health

44, 101-105. G.T.,

Dickinson,

high-frequency

hearing

J., Howd,

R.A.

loss in weanling

and Rebert, rats exposed

C.S. (1983) Transient to toluene.

cognitive

Neurobehav.

Toxicol.

deficits

and

Teratol.

5,

53-57. 10 Rebert,

C.S.,

Sorenson,

rats evidenced 11 Merigan,

W.H.,

in primates.

S.S., Howd,

by the brainstem Barkdoll,

Toxicol.

Appl.

R.A.

and Pryor,

auditory-evoked

E. and Maurissen, Pharmacol.

G.T.

response. J.P.J.

62, 342-345.

(1983) Toluene-induced Neurobehav.

Toxicol.

(1982) Acrylamide-induced

hearing Teratol.

loss in

5, 59-62.

visual impairment

340

12 Merigan, effects

W.H.,

Barkdoll,

on the macaque

E., Maurissen,

visual system.

J.P.J.,

Eskin, T.A. and Lapham,

I. Psychophysics

L.W. (1985) Acrylamide

and electrophysiology.

Invest.

Ophthalmol.

Visual Sci. 26, 309-316. 13 Eskin, T.A., macaque

Lapham,

visual

L.W.,

system.

Maurissen,

II.

J.P.J.

Retinogeniculate

and Merigan,

W.H.

morphology.

Invest.

(1985) Acrylamide Ophthalmol.

effects on the Visual

Sci. 26,

317-329. 14 Maurissen,

J.P.J.,

to acrylamide. 15 Schwann,

Weiss, B. and Davis, H.T.

Toxicol.

Appl.

Pharmacol.

T. (1839) Mikroskopische

dem Wachsthum

der Thiere

Untersuchungen

und Pflanzen,

Pathol.

F. (1822) Experiences

2, 276-219,

18 Miiller, 19 Weber,

E.H.

(1834)

E.H.

22 Green,

De

Pulsu,

G.T.

York,

J. Physiol.

Exp.

und der Thiere,

Auditu

Annotationes

Anatomicae

Huntington,

Hazards,

In: R. Wagner

(Ed.), Handworterbuch

Vol. 1 and 2, Breitkopf Theory

and man.

Science

der

und Hartel,

and Psychophysics,

Leipzig.

Robert

E. Krieger

127, 383-389.

a 1’Etude de la Medecine toxicology

and I. de Groot

HEW

et

NY.

C. (1856) Introduction

B.L. Johnson

et Tactu.

pp. 481-588.

der Psychophysik,

Publication

Washington, DC. 26 Guilford, J.P. (1954) Psychometric 27 Corso,

London.

des Menschen

(1974) Signal Detection

Paris, 1966, p. 212. 25 Stokinger, H.E. (1974) Behavioral tional

and Preston,

des nerfs rachidiens.

des Gesichtssinnes

und das Gemeingeftihl.

S.S. (1958) Measurement

Xintaras,

Strahan

des racines

Physiologie

Braunschweig,

(1860) Elemente

Co.,

24 Bernard,

of the Brain,

Resorptione,

(1846) Der Tastsinn

D.M. and Swets, J.A.

23 Stevens,

und

Lipsiae.

Vol. 3, Vieweg,

Publishing

in der Struktur

366-371.

C.F. Koehler,

Physiologie, 21 Fechner,

exposed

Leipzig.

Physiologicae, 20 Weber,

in monkeys

Berlin.

sur les fonctions

J. (1826) Zur vergleichenden

C. Cnobloch,

thresholds

tiber die Uebereinstimmung

Sander,

16 Bell, C. (1811) Idea of a New Anatomy 17 Magendie,

(1983) Somatosensory

71, 266-279.

in the development

(Eds.),

No.

Behavioral

(NIOSH)

Methods,

J.F. (1967) The Experimental

Experimentale,

McGraw-Hill

Psychology

of threshold

Toxicology.

74-126,

Editions

US

Government

Behavior,

Beltond,

limit values.

Early Detection

Book Company,

of Sensory

Pierre

In: C.

of Occupa-

Printing

New York,

Office, NY.

Rinehart

and Winston,

Psychophysics,

and

New

NY.

28 D’Amato,

M.R.

McGraw-Hill 29 Scharf,

(1970)

Experimental

Book Company,

B. (1975) Experimental

30 Gescheider, Hillsdale,

G.A.

(1976)

Psychology:

New York, Sensory

Psychology,

Psychophysics.

Methods

32 Cornsweet,

T.N.

33 Brownlee,

K.A.,

samples.

Scott, and

Foresman Theory,

Hodges,

and Co.,

Lawrence

Glenview,

Erlbaum

IL.

Associates,

Stat. Psychol.

H. (1965) Sequential

J. Psychol.

75, 485-491. method

with small

estimation

of points

on a psychometric

efficient

estimates

on probability

function.

18, l-10.

and Creelman,

Sot.

Am. 41, 782-787.

J.M.

Am.

M. (1953) The up-and-down

48, 262-277.

M.M.

36 Findlay,

35, 41 l-422.

in psychophysics.

Jr., J.L. and Rosenblatt,

G.B. and Levitt,

Br. J. Math.

Acta Otolaryngol.

(1962) The staircase-method

J. Am. Stat. Assoc.

34 Wetherill,

Acoust.

Learning,

NJ.

31 BCkCsy, G. von (1947) A new audiometer.

35 Taylor,

Methodology,

NY.

C.D.

(1978) Estimates

phys. 23, 181-185. 37 Pentland, A. (1980) Maximum

(1967) PEST:

on probability likelihood

functions: estimation:

a more virulent the best PEST.

PEST. Percept.

functions.

Percept.

Psycho-

Psychophys.

377-379. 38 Shelton,

B.R.,

procedures.

Picardi,

J. Acoust.

M.C. and Green, Sot.

D.M. (1982) Comparison

Am. 71, 1527-1533.

of three adaptive

J.

psychophysical

28,

341

39 Watson,

A.B.

and Pelli, D.G.

Psychophys. 40 Mad&an,

R.

alternative 41 Relkin,

E.M.

Williams,

and Pelli, D.G.

43 Blackwell,

H.R.

44 Lukaszewski,

adaptive

(1987) Probe

psychometric

psychometric Percept.

tone thresholds

J. Acoust.

Sot. Am.

discrimination:

functions.

method.

Percept.

procedures

Psychophys.

in the auditory

in two-

42, 240-249.

nerve measured

by two-

82, 1679-1691.

effects of probability

M.,

of stimulus

presentation.

J. Exp.

Stamm,

J.S.

J. Acoust.

and

Pribram,

lesions in monkeys.

R.A.

(1963) Detection

procedures

upon

psychophysical

measure-

44, 306-315.

D.N. (1962) Auditory

and motivation.

parietal

of data collection

J. Exp. Psychol.

J.S. and Elliott,

46 Campbell,

Maximum-likelihood

and recommendations.

(1952) The influence

feedback,

posterior

a Bayesian

18, 465-469.

ment of two sensory

45 Wilson,

(1987)

procedures.

T.F. (1972) Duration

Behav.

nique,

D.

evaluation

forced-choice

42 Elsmore, Anal.

and

forced-choice:

interval

(1983) QUEST:

33, 113-120.

threshold

Sot. Am.

K.H.

(1960)

J. Comp.

Deficits

Physiol.

of a noise signal

as a function

of forced-choice

tech-

discrimination

after

34, 223-228. in roughness

Psychol.

of varying

53, 535-539.

duration.

J. Acoust.

Sot.

Am.

35,

1732-1737. 47 Rose,

R.M.,

Percept.

Teller,

D.Y.

Psychophys.

48 Yager,

D. and Thorpe,

Animal

and

Rendleman,

P. (1970) Statistical

Psychophysics:

S. (1970) Investigations

of goldfish

The Design and Conduct

of Sensory

color vision.

50 Dyck,

functions.

P.J.,

(1978)

cutaneous

I.R.,

in man.

Hillsdale,

54 Pasternak,

T., Flood, W.H.

D.G.,

Eskin, T.A.

cells. Vision

I.P.

(1927) Conditioned

57 Northmore,

for measurement

of hearing.

P.E.,

Karnes,

touch-pressure,

J. and Bushek,

vibration,

and

W.

thermal

4, 502-510.

(translated

D.P.M.

analysis.

pathway

and Eskin,

Cortex

(Ed.),

In: M.S. Mayzner Research, Lawrence

and T.R. Erlbaum

of Sensation

and Perception,

Addison-Wesley

MA, p. 17.

ganglion

Cerebral

Neurol.

Ness, A., Caskey, to evaluate

J.M. (1981) Fundamentals

of probit

in the cat retinogeniculate

56 Pavlov,

procedure

Klein, S.A. and Teller, D.Y. (1985) Statistical

implications

retinal

P.C.,

systems

Ann.

Reading,

functions:

55 Merigan,

O’Brien,

Stebbins

Appleton-Century-Crofts,

NJ, pp. 100, 115.

Company,

53 McKee, S.P.,

estimates.

Am. 57, 431-436.

(1978) Psychophysical methods and the minicomputer. Minicomputers in Sensory and Information-Processing

M.W. and Shefner,

Publishing

Sot.

of automated

sensation

Associates, 52 Levine,

J. Acoust

Zimmerman,

Introduction

51 Penner, M.J. Dolan (Eds.),

of staircase In: W.C.

Experiments,

New York, NY, pp. 259-275. 49 Pfingst, B.E., Hienz, R. and Miller, J. (1975) Reaction-time II. Threshold

properties

8, 199-204.

Percept.

T.A.

properties

Psychophys.

and Merigan,

W.H.

by 2,5-hexanedione.

of forced-choice

(1985) Selective damage

J. Neurosci.

(1986) Spatio-temporal

psychometric

37, 286-298. to large cells

5, 1641-1652.

vision of macaques

with severe loss of Pp

Res. 26, 1751-1761. Reflexes:

An Investigation

by G.V. Anrep),

and Muntz,

(Scardinius erythrophthalmus),

W.R.A.

Dover

(1974) Effects

measured

of the Physiological

Publications,

New York,

of stimulus

with a classical

Activity

size on spectral

conditioning

of the

NY, 1960.

paradigm.

sensitivity Vision

in fish Res. 14,

503-514. 58 Otis, L.S., Cerf, in the goldfish. 59 Passe,

D.H.

pigeon.

J.H.

autonomic Crofts,

B.F.

Behav.

as a psychophysical

J. Comp. NY.

of auditory

Physiol.

(1938) The Behavior

New York,

(1957) Conditioned

inhibition

of respiration

and heart rate

paradigm:

absolute

visual

sensitivity

in the

in the rat by conditioning

of an

36, 133-139.

(1951) Measurement

response.

G.J.

126, 263-264.

(1981) Autoshaping

J. Exp. Anal.

60 Jamison, 61 Skinner,

J.A. and Thomas, Science

intensity

Psychol.

of Organisms.

thresholds

44, 118-125. An Experimental

Analysis,

Appleton-Century-

342

62 Terman,

M. and Terman,

operant-psychophysical 63 Clack,

T.D.

measuring

and

J.S.

(1972) Concurrent

test. Percept. Herman,

auditory

P.N.

thresholds

64 Estes, W.K. and Skinner,

variation

Psychophys.

(1963)

A single-lever

in the monkey.

of response

bias and sensitivity

in an

11, 428-432. psychophysical

J. Audit.

adjustment

procedure

for

Res. 3, 175-183.

B.F. (1941) Some quantitative

properties

of anxiety.

J. Exp. Psychol.

29,

390-400. 65 Sidman,

M., Ray, B.A.,

mutant 66 Smith, Crofts,

D. and

(Eds.),

P. (1977) Animal

(Ed.)

Behavior,

(1970)

Animal

P.M.

nonhuman

and

Young,

subjects.

Press,

70 Hoffman,

(1966) Hearing

technique.

of Sensory

New York,

New York,

J.S.

In: A.W.

Psychophysics.

Prentice-Hall,

In: W.C. Stebbins

Experiments,

(1985)

Hayes

In: W.K.

Englewood

Appleton-Century-

The

Psychophysical

(Ed.),

Honig

Cliffs,

Design

and

J.E.R.

Staddon

NJ, pp. 514-539.

and

Conduct

of Sensory

Ex-

dysfunction

in

NY. assessment

Toxicology

of

sensory

of the Eye, Ear, and Other

Special

Senses,

pp. 79-90.

H.S. and Ison, J.R. (1980) Reflex modification

findings

and vision in neurological

16, 377-402.

psychophysical

Psychophysics:

Appleton-Century-Crofts,

69 Blough,

as an animal

The Design and Conduct

of Operant

W.C.

periments,

J.M.

Exp. Neurol.

NY, pp. 125-159.

Blough,

Handbook

68 Stebbins,

suppression

Psychophysics:

New York,

67 Blough,

R.L. and Klinger,

for their evaluation.

J. (1970) Conditioned

(Ed.), Animal

Raven

Sidman,

mice: a method

and their implications

for how the nervous

in the domain

system processes

of startle.

sensory

I. Some empirical

input.

Psych.

Rev. 87,

175-189. 71 Sechenov,

I.M. (1863) Reflexes of the Brain (translated

by S. Belsky), M.I.T.

Press, Cambridge,

MA,

1965, p. 14. 72 Yerkes,

R.M.

73 Young,

J.S. and Fechter,

(1905) The sense of hearing

nique for assessing 74 Kelly,

J.B.

Psychol.

and

L.D.

ototoxicity.

Masterton,

J. Comp.

J. Acoust.

Neurol.

procedures

Psychol.

for animal

15, 279-304. audiometry:

a tech-

Sot. Am. 73, 1686-1693.

B. (1977) Auditory

sensitivity

of the albino

rat.

J. Comp.

Physiol.

91, 930-936.

75 Prosen,

C.A.,

Petersen,

kanamycin-induced Acoust.

Sot.

M.R.,

hearing

of a robust

77 Hoffman, of Startle

procedures Behavior,

L.D. and Young, Johns

Hopkins

79 Wilber,

L.A.

(1979)

(Ed.),

81 Arezao, 82 Sosenko, functional

factors

in the behavioral

Publishing

thresholds

reinforcement

and

procedure.

J.

Threshold

Hearing

Press,

measures.

II. The anomalous

of startle.

The use of reflex

(Ed.), Neural

Mechanisms

NY, pp. 267-285.

In: 2. Annau

(Ed.), Neurobehavioral

Tox-

MD, pp. 23-42.

methods

University

analysis

In: R.C. Eaton

New York,

Baltimore,

measurement

Assessment,

Schaumburg, assessment

J.M.,

Corporation,

of startle.

Bull. 94, 3-17.

of threshold.

J.S. (1986) Reflexive University

Gadia,

testing

H.H.

Park

and Press,

special

considerations.

Baltimore,

In:

W.F.

MD.

M.T.,

Stat.

C. (1986) Thermal

sense in diabetic

Natori,

for the detection

Br. J. Math.

and of

27, 505-507.

and Laudadio,

of thermal

1741-1744. 83 Kershaw, C.D. (1985) Statistical periments.

(1978) Auditory

in the domain

Psychol.

Eur. J. Clin. Pharmacol.

J.C.,

quantitative

W.C.

by a positive

D., de Weisse, C., Meyer, J.J., Bousquet, A. Safran, A.B., Desangles, A. (1984) Evaluation of visual function in healthy subjects after administration

80 Lupolover, Y., Assimacopoulos, Ro 15-1788.

pig assessed

phenomenon.

and the assessment

Plenum

icology,

Rintelmann

D.B. and Stebbins,

H.S. (1983) Reflex modification

and ubiquitous

H.S. (1984) Methodological

modification 78 Fechter,

Moody,

loss in the guinea

Am. 63, 559-566.

76 Ison, J.R. and Hoffman, history

in frogs.

(1983) Reflex inhibition

N., Ayyar,

D.R.,

of diabetic

peripheral

properties Psychol.

neuropathy.

of staircase

38, 35-43.

Ramos,

tester.

Device

for

35, 590-592.

L.B. and Skyler,

neuropathy.

estimates

sensitivity

Diabetes

Arch.

from two interval

J.S. (1987) NeuroIntern. forced

Med.

147,

choice ex-

343

84 Dixon, Stat.

W.J.

and Mood,

Assoc.

85 Arezzo,

J.C.

technique

A.M.

(1948) A method

for obtaining

and analyzing

sensitivity

data.

J. Am.

device.

A new

43, 109-126. and Schaumburg,

for detecting

H.H.

sensory

(1980) The use of the Optacon

loss in individuals

exposed

as a screening

to neurotoxins.

J. Occup.

Med.

22,

461-464. 86 LeQuesne, lesions.

P.M. and Fowler,

J. Neurol.

87 Kardel,

T. and Nielsen,

Acta Neurol. 88 Bleecker,

Stand.

M.L.

Cassito

Press,

Elmsford,

90 Arezzo, titative 91 Anger, Setzer,

and

W.K.,

fumigators

Moody,

clinical

H.H. H.H.

D.J.,

P.H. sensory

94 Nielsen, vibratory 95 Bleecker, vibration Biological Service, 96 Onofrj, rat VEP:

and

L., Burg,

and electrophysiological neuropathies.

in Occupational

foot study.

In: R. Gilioli,

Health,

Pergamon

Galer,

B.S.,

Dutcher,

and Schaumburg, neuropathy

perception

Soffar,

induced

is prevalent

in patients

and Moreland,

R. (1984) Carpal

Monitoring Springfield, M., Harnois, a comparison

for the Effects

R.K.,

C. and Bodis-Wollner, of flash and pattern

Russo,

J.M.,

Berger,

Screening

pesticide

Toxicologist

7, 134.

A., Arezzo,

sensory

testing

Arch.

Dickerson,

N.,

of soil and structural

7, 137-156.

(1987)

(OPIDN).

in chronic

Mizruchi,

demonstrates

Neurol.

renal failure.

J.C.,

plant M.,

that sub-

44, 944-946.

IV. An analysis

of the

191, 287-296. tunnel

studies.

of Exposure

VA, No. PB86-242641,

peripheral

a simple device for quan-

evaluation

with cancer.

nerve function

for

l), 169.

B.J.,

D.A.

neuropathy

Portenoy,

electrodiagnostic

screening

Neurotoxicology

Otto,

(1987) Quantitative

Acta Med. Stand.

versus

Taylor,

fluoride. A. and

threshold.

threshold

35 (Suppl.

W.S.,

delayed

J.P., H.H.

Rapid

K. (1986) Neurobehavioral

and sulfuryl

S.A.,

(1983)

33, 626-629.

C. (1985) The Vibratron:

J., Brightwell,

bromide

Soliman,

C.A.

Neurology

sense. Neurology

V.K. (1972) The peripheral M.L.

Methods

Petersen,

and Laudadio,

B.L. and Hicks,

using methyl

R.B.,

Wiernik,

A clinical

device for peripheral

Neurobehavioral

with the Optacon.

for organophosphorus

93 Lipton,

neuropathy.

a new screening

of tactile/vibratory

Johnson,

92 Svendsgaard,

-

V. FOB (Eds.),

Schaumburg,

evaluation

workers

V.K. (1974) Hepatic

Schaumburg,

J.V.,

with neuropathic

50, 513-526.

a field study

J.C.,

in diabetics

49, 1191-1194.

NY, pp. 41-46.

J.C.,

neuropathy:

Psychiatry

(1983) Optacon

M.G. 89 Arezzo,

C.J. (1986) A study of pain threshold

Neurosurg.

syndrome:

comparisons

In: Conference

in the Workplace,

of alterations

on Medical

National

Screening

Technical

in and

Information

p. 27. I. (1985) The hemispheric stimulation.

Exp. Brain

distribution

of the transient

Res. 59, 427-433.