Patent No. 5406956 Method and apparatus for truth detection

 

Patent No. 5406956

Method and apparatus for truth detection (Farwell, Apr 18, 1995)

Abstract

A method of detecting information stored in the brain of a subject includes presenting to the subject in oddball series Probe, Target, and Irrelevant stimuli. The Probe stimuli are relevant to a situation under investigation; the Irrelevant stimuli are not; and the Target stimuli are identified to the subject as being noteworthy, and in response to which the subject is instructed to perform a task. The Target stimuli like the Probe stimuli are relevant to the situation under investigation. The method also includes detecting electrical brain responses for each of the stimuli; analyzing the responses for uncovering an event related brain potential; and comparing the Probe responses with the Target responses to determine whether the subject recognizes the Probes, and comparing the Probe responses with the Irrelevant responses to determine whether the subject does not recognize the Probes. Three exemplary headbands are disclosed for positioning electrodes at preferred locations on the subject's scalp for obtaining electrical responses therefrom.

Notes:

INTRODUCTION

Conventional Psychophysiological Detection of Concealed Information

Attempts to discover concealed information stored in the brain of another human being are as old as human history. The purpose of this report is to propose a new paradigm for the psychophysiological detection of concealed information, to test the new paradigm, to contrast it with the conventional paradigm, and to examine its theoretical and practical implications.

When an individual has a particular significant experience, information regarding that experience is stored in the brain. The task at hand for another person lacking knowledge of the original event who wishes to discover the facts regarding the event is to detect this concealed information stored in the brain.

For millennia, the method for achieving this end has been interrogation. Interrogation consists of asking the person questions regarding the information of interest, and attempting to motivate him/her to reveal the information (Aubry & Caputo, 1980). The Farwell Truth Detector introduced here detects the concealed information in a different manner: by measuring electrical brain activity that results, in certain circumstances, from the presence of the concealed information.

In conventional detection of concealed information, when an individual is questioned regarding information that he or she may be motivated to conceal, some means of assessing the credibility of the statements made must be practiced in order for the questioning to be productive (Richardson, 1963). Credibility assessment can be achieved in two ways: 1) through observation of overt behavior, and 2) through observation of non-overt behavior through psychophysiological measurement.

Although the detection of concealed information and credibility assessment through psychophysiological means have been practiced successfully for thousands of years, the utilization of scientific instrumentation in this pursuit has been practiced for only about a century.

This practice is generally known as lie detection, detection of deception, or polygraphy. What is actually detected, however, is neither lies nor deception; and there are many other uses of the polygraph. Thus, the term "interrogative polygraphy" is perhaps a more suitable term. I shall use the term "psychophysiological detection of concealed information" as inclusive of both interrogative polygraphy and alternative methods of detection involving the observation of psychophysiological responses without instrumentation.

Prior to the research reported here, interrogative polygraphy and all other recorded attempts to utilize psychophysiological measures in the service of detecting concealed information were based on an emotional/physiological arousal paradigm (Munsterberg, 1908; Reid & Inbau, 1977; Matte, 1980; Ansley, 1975). It is well known that when faced with a potentially threatening situation, the body prepares itself to become vigorously, physically active in order to deal with that situation. This preparation takes place when strong emotions are aroused, even if an actual physical response of fighting or fleeing is known to be out of the question. Psychophysiological methods of detecting concealed information have been founded on eliciting such emotions (Ferguson & Miller, 1973; Gugas, 1979).

The fundamental procedure that has been developed based on this phenomenon combines interrogation and psychophysiological credibility assessment (Inbau & Reid, 1953; 1962; Aubry & Caputo, 1980). There are three aspects to this procedure:

1) The subject is questioned regarding the event in question, and his/her answers are noted.

2) The subject is confronted with questions or statements regarding the event in question, in a context that is potentially threatening or fear inducing. According to the theory, a subject who has actual information that he has not accurately revealed (a "guilty" subject) will experience greater negative emotions than a subject who is not concealing information (an "innocent" subject).

3) According to the theory, this negative emotion will result in a physiological activation preparatory for strenuous physical activity that can be measured through various channels--primarily perspiration on the palms, changes in cardiovascular activity, and changes in breathing patterns (Reid & Inbau, 1977; Matte, 1980; Ferguson & Miller, 1974). (Several other responses, including salivation, peripheral blood flow, and pupillary responses, have also been measured.)

What is actually measured is the physiological response. According to the theory of conventional interrogative polygraphy, a large physiological response indicates a large emotional response, which in turn indicates guilt or deception (Ferguson & Miller, 1973; Matte, 1980). As will be described in detail in subsequent sections, this simple theory has remained essentially unchanged for two thousand years (Ferguson & Miller, 1974). This is not to say that there has been no progress in interrogative polygraphy during that time. The progress that has taken place--and, particularly during the last quarter century, this has been extensive--has been of an empirical nature. As a recent report prepared under the auspices of the Department of Defense (Department of Defense, 1984) concluded, "The polygraph field is one of those rare situations where the practice has outpaced the research."

Polygraphers have had extensive opportunities to develop a database consisting of physiological responses on the one hand and confessions or other determinations of guilt or innocence on the other. An extensive and systematic body of knowledge has developed regarding the specific patterns of responses that correspond with an ultimate determination that the subject was guilty or innocent (Munsterberg, 1908; Marston, 1917; Larson, 1922, 1932, 1969; Keeler, 1930; Reid & Inbau, 1977; Raskin, 1989). In parallel with this, much has been learned about criminal psychology and the interrogative techniques that are most effective in eliciting revealing responses, both overt and psychophysiological (Munsterberg, 1908; Inbau & Reid, 1962; Gugas, 1979; Aubry & Caputo, 1980).

A simple theory provided the original impetus to make physiological measurements. For the purpose of practical results in investigations, the theoretical foundations, theoretical implications, and intervening variables are not of primary importance (Department of Defense, 1984). What has been systematically and extensively investigated is 1) what physiological patterns can be taken as reliable indicators of guilt, and 2) what techniques of interrogation are most effective in making use of these responses (Reid & Inbau, 1977; Aubry & Caputo, 1980).

In order to detect information that is stored in the brain, conventional interrogative polygraphy resorts to interrogations to elicit true or false statements, and psychophysiological credibility assessment or lie detection to determine whether the subject can be believed or not. The lie detection procedure involves presenting a stimulus that is relevant to the concealed information in a potentially threatening or upsetting situation and measuring physiological responses--on the theory that a guilty individual will emit a greater emotional response to the relevant stimuli, and this greater emotional response will be reflected in a greater physiological response (Munsterberg, 1908; Marston, 1917; Selling, 1938; Ansley, 1975; Matte, 1980; Gale, 1988, Raskin, 1989).

The theoretical basis, then, of conventional psychophysiological detection of concealed information is a general arousal theory. Since the origin of interrogative polygraphy, psychophysiological research has gone beyond a general theory of arousal to explore a variety of more specific psychophysiological responses related to emotions (Lacey, 1967; Obrist, Webb, Sutterer, & Howard, 1970; Schwartz, Weinberger, & Singer, 1981; Mathews, May, Mogg, & Eysenck, 1990; Davidson, Ekman, Saron, Senulis, & Friesen, 1990; Ekman, Davidson, & Friesen, 1990; Lang, Bradley, & Cuthbert, 1990; Eysenck, Mogg, May, Richards, & Mathews, 1991). This further development, however, has not fundamentally changed the theory of interrogative polygraphy. Progress in conventional interrogative polygraphy has consisted primarily of an empirical investigation of what psychophysiological responses are indicative of guilt or deception. A variety of different patterns have been shown to be reliable predictors in this regard (Reid & Inbau, 1977; Matte, 1980; Ansley, 1975).

Two closely related areas where considerable theoretical, as well as practical, advances have taken place in recent years are criminal assessment through observation of overt behavior, and criminal psychology and interrogation (Richardson, 1963; Inbau & Reid, 1962; Gugas, 1979; Aubry & Caputo, 1980). These are beyond the scope of this paper.

Remaining within the realm of general arousal theory, two different methods of interrogative polygraphy employing brain waves could be designed. Alpha waves have been hypothesized to be an indicator of cortical deactivation and general relaxation, and both alpha blocking and beta waves have been hypothesized to be an indicator of cortical activation. Although these hypotheses are not universally accepted, they have received a considerable measure of empirical support. (See, for example, Nowlis & Kamiya, 1970; Brown, 1971; Andersen & Andersson, 1968; Lippold & Novotny, 1970; Shaw, Foley, & Blowers, 1970; Plotkin & Cohen, 1976; Travis, Kondo, & Knott, 1975; Lynch & Paskewitz, 1971; Davidson, 1984a, 1984b; 1987; Kinsbourne & Bemporad, 1984; Leventhal & Tomarken, 1986; Silberman & Weingartner, 1986; Tucker & Frederick, 1989; Davidson & Tomarken, 1989; Davidson, Ekman, Saron, Senulis, & Friesen, 1990). Based on this, one could reasonably advance the hypothesis that the presentation of a crime-relevant stimulus would elicit greater alpha blocking and/or greater beta activity in a guilty person than in an innocent person.

Another theory holds that voltage potential at the scalp is an indication of the level of activation of the brain. There is strong evidence that this is the case in certain limited circumstances, particularly situations involving motor potentials (Kornhuber & Deeke, 1965; Kutas & Donchin, 1980). Some have theorized that slow event-related brain potentials are the result of modulations in the level of widespread cortical activation: negative potentials indicate activation; apparently positive potentials can be explained as transient deactivation (Desmedt, 1980; Verleger, 1988). If this theory is correct, then one could on this basis predict increased cortical negativity in a guilty subject's response to crime-relevant stimuli, since these stimuli would, as indicated by other measures, result in a higher level of activation. As will be described in detail in subsequent sections, the findings predicted in this paper are incompatible with this theory and this predicted result.

The New Paradigm

This paper proposes a new paradigm regarding concealed information and its relationship to psychophysiological measurements, and describes a series of experiments implementing and evaluating this paradigm. The paradigm is distinct from the emotional/physiological arousal paradigm, based on a general arousal theory, that has formed the basis of interrogative polygraphy to date. If the results are as predicted, this new paradigm promises to provide a new means of psychophysiological detection of concealed information that detects the presence of the information more directly and reliably than the conventional methods--without recourse to the mandatory elicitation of negative emotions or dependence on the linkage between negative emotions and physiological arousal.

The P300 Component

The theory and research presented here are based on the P300 component of the event-related brain potential, also referred to as P3 and P3b. There is considerable evidence that P300s are elicited by relatively rare stimuli within a series of more frequent stimuli, when--and only when--these rare stimuli provide information needed by a subject for the performance of a task assigned by the experimenter. For example, if a subject is instructed to count low tones in a series of 20% low and 80% high tones, the low tones will elicit a P300. If the subject is asked to ignore the tones and solve a mental problem instead, the same tones do not elicit a P300 (Duncan-Johnson & Donchin, 1977). If a series of photographs is presented containing a majority of neutral photographs along with a few photographs of politicians and a few photographs of movie stars, a subject will emit P300s in response to one but not the other of these categories, depending on which category he or she is instructed to count (Towle, Heuer, & Donchin, 1980). P300's, then, have been shown to be elicited by stimuli that are rare and explicitly task-relevant.

There are two major classes of theories to explain this phenomenon. One class of theories, exemplified by the context updating theory (Donchin, 1981; Donchin, Karis, Bashore, Coles, & Gratton, 1986; Donchin & Coles, 1988a; 1988b), holds that the P300 is the manifestation of an active information processing "subroutine" in the brain. According to context updating, when the explicitly task-relevant stimulus arrives, the subject updates his internal representation of the environment, and it is this active process that is manifested in a P300 on the scalp.

Another class of theories of the functional significance of P300 holds that the P300 is the result of a transient deactivation of the cortex (Desmedt, 1980; Verleger, 1988). The cortex has been activated in anticipation of the arrival of the task-relevant stimulus. When the stimulus arrives "context closure" occurs, the subject momentarily relaxes his vigilance, and the cortex is deactivated, resulting in an apparent positivity. Then the cortex is again activated in anticipation of the next stimulus, resulting in a return to the previous state of cortical negativity (that appears to be a return to baseline).

Context updating, then, holds the P300 to be a manifestation of the activation of a specific, information-processing process. Brain deactivation/closure holds the P300 to be the result of a transient cortical deactivation. Note that these two theories are not directly comparable. Context updating specifies the functional significance of P300 in terms of a specific information-processing function; it does not specify particular brain structures or physiological brain functions. Brain deactivation specifies particular electrophysiological events in particular brain areas--specifically, deactivation of certain cortical areas--and explicitly denies the existence of any specific, concomitant information-processing process. These differences will be discussed in detail in subsequent sections.

Psychophysiological Detection of Concealed Information with the P300

If the context updating theory is correct, it may be possible to design an interrogative polygraphy technique that takes advantage of the specific nature of the information-processing process manifested in P300 to draw conclusions regarding the information that is processed--and therefore possessed--by the subject. This is the perspective taken in the research described here.

On the other hand, if the cortical activation/deactivation theory is valid, it may be possible to develop a method of interrogative polygraphy through the elicitation of increased cortical activation and corresponding negativity by the presentation of crime-relevant stimuli, or through other interrogative techniques eliciting cortical activation.

The different predictions made by these two theories, and the practical applications for psychophysiological detection of concealed information, will be discussed in detail in subsequent sections.

It is well established that P300's are elicited by stimuli that provide information necessary for the performance of an explicit task assigned by the experimenter (Sutton, Braren, Zubin, & John, 1965). In previous research, P300's have been absent in the absence of relevance to such an explicit task (Sutton, Tueting, Zubin, & John, 1967; Duncan-Johnson & Donchin, 1977).

The hypothesis advanced here is that stimuli that are not explicitly task-relevant will nevertheless elicit a P300 if they are particularly significant to the subject due to his past experience with the subject matter of the stimuli. This is based on the context updating model. If this model is correct, then when a stimulus that is significant for the individual arrives., he or she can be expected to take particular note of it, thus revising his/her internal representation of the current environment and emitting a P300. If this is shown to be the case, this will lend support to the notion that the P300 is indeed the manifestation of a particular, active process. Moreover, it will demonstrate that this is a process that takes place upon the arrival of relevant information even when the subject is not instructed or motivated to engage in this process--in fact, when it is to the subject's disadvantage to do so.

Again, previous research has indicated that stimuli will elicit a P300 only when they are explicitly task-relevant, and not when they are not explicitly task-relevant even if they are recognized by the subject (e.g., movie stars when one is counting politicians; Towle, Heuer, and Donchin, 1980). A hypothesis advanced here is that explicit task relevance is not a necessary condition for the elicitation of a P300. What could be called "implicit task relevance"--a high measure of significance for the subject--is a sufficient condition on the task relevance dimension.

This paper sets forth a series of experiments to test this hypothesis, and also to investigate further the nature and prerequisites for implicit task relevance or significance.

The Research Program

The original research program comprises three experiments. (Experiments 1 and 2 were reported by Farwell & Donchin, 1986, 1991. Experiment 3 was reported by Farwell, 1992.) In Experiment 1, a particular set of stimuli was made implicitly relevant to a subject by requiring the subject to learn them through an interactive computer program and to carry out a mock espionage scenario utilizing the relevant information. On the following day, subjects were presented with the implicitly relevant "probe" stimuli interspersed with two other types of stimuli: irrelevant stimuli and target stimuli. All stimuli were visually presented short phrases. Subjects were asked to memorize a list of targets, and to press one button in response to targets and another button in response to all others. Target and probe stimulus probability were 0.17; irrelevant probability were 0.66. Each subject was tested with two different sets of stimuli, one containing probes relevant to an espionage scenario he or she has carried out ("guilty" condition), and one with probes associated with another scenario unknown to the subject ("innocent" condition).

The innocent condition, then, was simply a standard experiment of the kind that has been shown to produce P300's. Since the subject cannot distinguish the probes from the irrelevants, he sees only two kinds of stimuli: rare and task-relevant targets, and frequent irrelevants. A P300 is expected in response to the targets and not in response to the irrelevants and the probes (which are not recognized as such).

The guilty condition provides a test of the implicit task relevance hypothesis and indirectly of the context updating model on which it is based. If my hypothesis is correct, the subject will display large P300's to probes as well as to targets.

Experiment 2 sought to delineate further the nature of implicit task relevance. In Experiment 1, probe stimuli are learned only one day prior to testing, and are learned in the context of the experiment. In Experiment 2, the probe stimuli consisted of phrases relevant to an actual crime or socially undesirable act committed by the subject up to several years before testing. Stimulus sets were developed through discussion with the subject on the day prior to testing. In other regards, experimental design and predicted results were similar to Experiment 1.

Even if the results of Experiment 2 are as predicted, it is still possible that the significance of the probes is a function of the discussions that take place shortly prior to testing. Experiment 3 was designed to test the hypothesis that personal relevance or implicit task relevance can be achieved on the basis of significant life events alone, outside of the experimental setting, and that it can be retained over a period of years. Experimental procedure was as before, except that all information regarding the personal life events that give rise to the probe stimuli will be supplied by a third party who knows the subject well. Subjects will not know what stimuli will be presented or what events will be investigated prior to the test session. As before, the predicted result for a guilty subject is large P300's to targets and probes and not to irrelevants. If the results of Experiment 3 are as predicted, this will serve to delineate further the nature and time course of implicit task relevance, and provide further support for the context updating model of the functional significance of the P300. This will be discussed in detail in the section on "Hypotheses and Theoretical Significance of this Research."

The permission of each subject will be obtained for the acquisition of information from each specific informant. Any information so obtained will remain confidential.

These three experiments, if results are as predicted, can serve as the theoretical and practical foundation of a new form of psychophysiological detection of concealed information. All previous methods involve interrogation--asking the subject about the situation under investigation--and credibility assessment--attempts to determine if the subject is telling the truth (Reid & Inbau, 1977; Matte, 1980; Department of Defense, 1984; Aubry and Caputo, 1980; Raskin, 1986; Lykken, 1978). Credibility assessment in turn involves measurement of physiological responses on the theory that large physiological responses indicate a large emotional response, which in turn indicates that the sought after, crime-relevant information is indeed stored in the brain (Ferguson & Miller, 1973).

The method proposed here seeks to reveal the crime-relevant information stored in the brain through measurement of the psychophysiological manifestations of information-processing brain activity. The Farwell Truth Detector described here does not depend on the elicitation of certain emotions or the measurement of the putative physiological correlates of emotions. The system creates a situation such that the presence of crime relevant information stored in the brain--regardless of emotions or arousal levels--will result in the implementation of a certain information-processing brain function that can be detected electrophysiologically.

One advantage of this approach is that both emotions and the accompanying autonomic nervous system arousal can be multiply determined, whereas the process manifested in the P300 is a quite specific one. On the emotional level, there may be many reasons other than guilt or falsehood why a person may be aroused (or fail to be aroused) by being questioned about a crime. Moreover, the autonomic nervous system is activated not only by emotions. It is responsible for maintaining homeostasis in the system as a whole. It continually readjusts the balance between anabolic and catabolic processes, maintains a suitable temperature in the different areas of the body, supplies nutrients and oxygen to different organs on demand and according to a complex system of shifting priorities, and makes a multitude of adjustments in different bodily systems in response to continually changing conditions in the internal and external environment. Thus, the changes in autonomic nervous system activity measured by conventional polygraphy may be the result of a multitude of psychological and physiological factors, some related to the subject of the investigation and some unrelated. Moreover, a knowledgeable subject may deliberately engage in psychological or physical activities that influence autonomic nervous system activity during an examination, and such countermeasures may have considerable effectiveness (Honts, Hodes, & Raskin, 1985; Honts, Raskin, & Kircher, 1987).

By contrast, there is strong evidence that the information-processing activity of which the P300 is a manifestation is a specific one (Donchin, Karis, Bashore, Coles, & Gratton, 1986). The brain-wave information detection (BID) system described here attempts to give the subject a task where that particular information-processing function comes into play if and only if the relevant information is stored in the brain. As will be described in detail in subsequent sections, the Farwell Truth Detector attempts in this way to establish a very specific link between a psychophysiological measurement and the presence or absence of specific, relevant information stored in the brain.

EARLY PSYCHOPHYSIOLOGICAL DETECTION OF CONCEALED INFORMATION

The Veda, the earliest record of human experience, contains the first record of systematic methods for the psychophysiological detection of concealed information. The Veda is said to be well over 10,000 years old. However, since it was passed down orally for many generations by pundits who memorized it syllable for syllable and taught it to the upcoming generation from an early age, it is difficult to ascertain its age accurately. It may have been first written down about 900 B.C. Ancient Vedic scientists used the pulse, measured from the wrist, to diagnose a wide variety of physical, mental, emotional, and spiritual imbalances, including both guilt over past deeds and tendencies to violence and other crime. The Mitakshara Shastra of the Ayur Veda, that section of the Vedic literature concerned with both psychophysiology and health, offered the following method for detecting individuals who had committed the particularly heinous crime of poisoning:

A person who gives poison may be recognized. He does not answer questions, or they are evasive answers; he speaks nonsense, rubs the great toe along the ground, and shivers; his face is discolored; he rubs the roots of the hair with his fingers; and he tries by every means to leave the house . . . " (Mitakshara Shastra. See Wise, 1845, p. 394; see also Trovillo, 1939; Horvath, 1973).

Like the methods employed in modern interrogation, this ancient account calls for the observation of both psychophysiological and overt behavioral symptoms.

The first practice of psychophysiological detection of concealed information recorded in detail was by the Greek physician and physiologist Erasistratos in the third century B.C. (Appian of Alexandria, early second century A.D., 1962; Plutarch, first century A.D./1952; Valerius Maximus, first century A.D./1888; see also Mosso, 1896; Trovillo, 1939; Horvath, 1973; Mesulam & Perry, 1972). Erasistratos used psychophysiological detection of concealed information, without the benefit of modern psychophysiological recording equipment, to solve a particularly difficult medical case. His patient was a young man named Antiochus, the son of Seleucus I of Syria, a former general of Alexander the Great who had recently married a beautiful younger woman named Stratonice. Antiochus fell in love with his new stepmother, and realizing the hopelessness of his situation, attempted to hide his passion. He soon became very ill.

Erasistratos concluded that the young man had no physical ailment, and consequently his illness must have an emotional origin. He made careful psychophysiological observations of Antiochus' responses to various stimuli in the environment in order to determine the cause of the illness. Suspecting that the young man's difficulties arose from his feelings for another individual, and that he was doing his best to avoid any overt behavior that would give him a way, Erasistratos made some of the same measurements that are now incorporated in modern polygraph testing, along with other related observations. Here, in the words of Plutarch (first century A.D./1952), is what Erasistratos observed:

He . . . waited continually in his chamber, and when any of the beauties of the court made their visit to the sick prince, he observed the emotions and alterations in the countenance of Antiochus, and watched for the changes which he knew to be indicative of the inward passions and inclinations of the soul. He took notice that the presence of other women produced no effect upon him; but when Stratonice came as she often did, alone or with Seleucus, to see him, he observed in him all of Sappho's famous symptoms,--his voice faltered, his face flushed up, his eyes glanced stealthily, a sudden sweat broke out on his skin, the beatings of his heart were irregular and violent, and, unable to support the excess of his passion, he would sink into a state of faintness, prostration, and pallor.

The "sudden sweat" described by Erasistratos constitutes the primary symptom recorded by modern interrogative polygraphy. The "irregular and violent beatings of the heart"--also a response recorded in modern lie detection--could only have been observable if Erasistratos had actually taken the pulse at the appropriate times (a fact that was pointed out by another great early Greek physician and scientist, Galen of Pergamum, second century A.D.).

Here Erasistratos exhibits a sophisticated theoretical understanding of psychophysiology, the same theoretical understanding that forms the basis of all modern conventional interrogative polygraphy prior to the system proposed here (Matte, 1980; Ferguson & Miller, 1973). Erasistratos must have realized that emotions and bodily functioning are intimately related. He recognized that a stimulus in the external world could produce a visceral response due to its particular significance for an individual. He replicated his findings on the relationship of the stimulus and the psychophysiological response through repeated observations at different times. Erasistratos also must have understood that the visceral responses could reveal emotions--and, by inference, the concealed information that gave rise to them--even if the subject was determined to reveal neither the emotion nor the facts giving rise to it through any overt means. Moreover, Erasistratos' scientific study was a controlled one. He assured himself of the specificity of the stimulus by comparing the subject's response to the critical stimulus to his responses to the presence of other individuals of the court.

Galen of Pergamum, often hailed as the father of modern medicine and an admirer of Erasistratos, further developed the theoretical and practical approach introduced by his predecessor (Mesulam & Perry, 1972). He applied it to solve a similar problem, this time with a woman suffering from insomnia. After examining her, he concluded that it was likely that her problem was due to "some trouble she was unwilling to confess." (Galen, second century A.D.)

While I was convinced the woman was afflicted by not bodily disease, but rather that some emotional trouble grieved her, it happened that at the very moment I was examining her this was confirmed. Someone returning from the theater mentioned he had seen Pylades dancing. Indeed, at that instant, her expression and the color of her face were greatly altered. Attentive, my hand laid on the woman's wrist, I observed her pulse was irregular, suddenly violently agitated, which points to a troubled mind. The same thing occurs in people engaged in an argument over a given subject.

The next day, I told one of my following that when I went to visit the woman he was to arrive a little later and mention that Morpheus was dancing that day. When this was done the patient's pulse was in no way changed. And likewise, on the following day, while I was attending her, the name of the third dancer was mentioned, and in like fashion the pulse was hardly affected at all. I investigated the matter for a fourth time in the evening. Studying the pulse and seeing that it was excited and irregular when mention was made that Pylades was dancing, I concluded that the lady was in love with Pylades and in the days following, this conclusion was confirmed exactly.

Galen's scientific application of psychophysiology to detect concealed information goes beyond that of Erasistratos. Rather than passively waiting for the arrival of the critical stimulus, Galen intentionally manipulated the environment to elicit the responses in question. He used control stimuli as well as relevant ones. Moreover, he recognized the concept, fundamental to modern interrogative polygraphy, of stimulus generalization: the name, rather than the individual, was enough to elicit the response (Mesulam & Perry, 1972). In a criticism of physicians ignorant of the psychophysiological connection between mind and body, Galen (second century A.D.) clearly describes the intimate relationship that forms the basis of modern conventional interrogative polygraphy.

Why did these things escape the notice of earlier physicians attending the lady described above?. They are arrived at by ordinary deduction, if the physician has even a meager knowledge of medicine.

Indeed, I think that it is because these physicians possess no clear conception of the ways the body tends to be influenced by the state of the mind. Perhaps it is because they do not even know that the pulse becomes turbulent because of strife and fears which suddenly disturb the mind.

In the tenth century the great Persian physician Avicenna, in his Canon medicinae, describes similar success to that of Erasistratos and Galen in diagnosing "love-sickness," and systematic identification of its object, in an individual who had been unwilling to confess his condition. Avicenna also expanded the concept of stimulus generalization to include not only the name of the relevant person but other related items such as her place of residence. Avicenna's (also known as Ibn Sina, tenth century A.D) description of his method is extremely similar to the conventional lie detection practiced by interrogative polygraphers today (Ibn Sina, 1608; Mesulam & Perry, 1972). In the following passage, Avicenna (Ibn Sina, 1608) describes the detection of a love-sick individual.

It is possible in this way to ascertain whom he loves, even when he will not reveal it himself . . . The nature of the cure is this: let several names be pronounced, repeating them many times, and place your finger on the patient's pulse. When it varies by a large fluctuation and then returns to normal, and this is repeated thereafter, and is put to the test many times, then the name of the one he loves will be known.

Again, similarly, make mention of her looks and habits and that in which she excels, her family, where she lives, so that any one of these things may be associated with the name of his loved one. Observe his pulse in such a way that when it fluctuates at the mention of one of these details the particular characteristics of his loved one may then be associated with a name and with an outstanding feature, by all of which she is to be recognized.

It is clear that the theoretical foundations of conventional interrogative polygraphy, and indeed a recognition of many of the critical psychophysiological responses to be measured, are founded on a two-thousand-year-old tradition. Modern conventional interrogative polygraphy has built upon this foundation with sophisticated instrumentation, and has developed a rich empirical data base.

THE DEVELOPMENT OF MODERN CONVENTIONAL INTERROGATIVE POLYGRAPHY

Interrogative polygraphy, the use of scientific instrumentation in psychophysiological detection of concealed information, began with Lombroso in the late nineteenth century (Lombroso, 1890, 1887, 1911, 1912; Ferraro, 1911). Lombroso made continuous recordings of cardiovascular activity by placing a subject's hand in a water-filled tank covered with a rubber membrane. Cardiovascular activity was reflected by changes in the volume of the fist, which modulated the water level. Changes in the water level were transferred by an air-filled tube to a revolving smoked drum. Lombroso's measurement met with some success in detecting actual criminals.

The first "polygraph" for recording cardiac activity that contained the essential features utilized in the modern polygraph for recording cardiovascular phenomena was developed by heart specialist Sir James Mackenzie in 1906 (Mackenzie, 1908; Gay, 1948). It was used, however, not for the psychophysiological detection of concealed information but for medical purposes. As early as 1908 Harvard professor Hugo Munsterberg (1908), an innovator in the psychophysiological study of emotion, proposed the use of all three of the measurements now commonly used in lie detection, along with several other behavioral and psychophysiological parameters. His description of the techniques is both insightful and poetic.

If a girl blushes when a boy's name is mentioned in the family sitting-room, we feel sure, even if she protests, that he is not quite indifferent to her young heart. If she opens a letter and grows pale while reading it, she may assure us that the event is unimportant; we know better . . .

Yes, the hidden feeling betrays itself often against the will . . . It may be easy to suppress intentionally the conspicuous movements by which we usually accentuate the emotions . . . But the lips and the hands and arms and legs, which are under our control, are never the only witnesses to the drama which goes on inside--if they keep silent, others will speak.

Our inspirations and expirations can be registered in finest detail and a variety of elegant methods are available. Perhaps the simplest "pneumograph" consists of a tube made of spiral wire and covered with rubber, to be attached by ribbons to the chest . . . As soon as such delicate methods of registration are applied, the intimate relation between feeling and breath becomes evident . . .

The same holds true for the heart beat, measured by the blood wave in the arteries; such a pulse writer is called a sphygmograph. It may be attached, for instance, to the wrist; a delicate lever presses against the wall of the blood vessel just where the finger of the physician would feel the pulse . . . When we write pulse and breathing together on the same drum, we see at once that even ordinary inspiration changes the pulse; while we inhale we have a pulse different from the pulse when we exhale. Far more influential are the feelings . . .

But there is still another way to observe the changes in our blood vessels. We may examine the quantity of blood, for instance, which streams to a limb, by means of the so-called plethysmograph . . . every emotional excitement speaks in the blood supply of every limb.

But we may go still further and point to expressions of emotion which are entirely beyond human senses. If we put our hands on two copper plates and make the weak galvanic current of a battery run through the plates and our body, we can, with the help of a delicate galvanometer, measure the slightest variations of the resistance of the current. Experiment shows that such changes occur, indeed, if our brain is excited; any emotional disturbance influences the resistance: it seems that the activity of the sweat-glands in the skin is under the nervous influence of our feelings, and the functioning of these glands alters the electrical conditions. A word we hear may excite us and at once the needle of the galvanometer becomes restless: there is no more uncanny betrayal of our inmost mind. (Munsterberg, 1908, p. 113)

A device incorporating all three of the measures included in a modern polygraph--pulse, breathing, and skin conductance--was proposed to a congressional hearing by Arthur McDonald in 1908, but such a device was not constructed until years later.

In 1914 Vittoria Benussi reported some success in detecting deception through measuring changes in breathing patterns (Benussi, 1914). In 1915 William Marston (Marston, 1917; 1938), a student of Munsterberg's and later also a professor at Harvard, began a series of experiments using blood pressure in the detection of deception. He used an ordinary sphygmomanometer to record blood pressure periodically during questioning. Marston also recorded breathing patterns and experimented with skin resistance measurements.

Burtt (1918) also experimented with respiratory measurements and blood pressure in lie detection. He held the blood pressure changes to be of greater diagnostic value.

The first continuous recording of blood pressure in interrogative polygraphy was accomplished by Larson (1921; 1922). He developed an instrument that simultaneously recorded blood pressure, pulse, and respiration continuously throughout a test session. Larson's device was used extensively in criminal investigations, and reportedly enjoyed a high degree of success (Larson; 1932).

In 1926 Keeler (1930) developed a machine that accomplished the same measurements as Larson's and incorporated a number of technical improvements.

In 1938 Keeler incorporated the third major component of the modern polygraph, namely the psychogalvanometer invented by Galvani in 1891. This provided for the measurement of the skin's resistance to (or conductance of) an electrical current, commonly known as electrodermal response (EDR), galvanic skin resistance (GSR), or skin conductance response (SCR).

By this time, considerable research had been published on skin resistance, but little of it had been related to the psychophysiological detection of concealed information. An exception was Summers (1939), who conducted thousands of laboratory experiments and investigated about 50 actual cases using galvanic skin resistance, reportedly achieving a high level of accuracy in detecting deception or guilt.

Another early researcher in the use of galvanic skin resistance in psychophysiological detection of concealed information was Wilson (Trovillo, 1939), who developed a new recording psychogalvanometer in 1930 and used it in collaboration with Keeler in several investigations. Shortly thereafter, the Chicago Police Crime Detection Lab first used galvanic skin resistance in conjunction with blood pressure measurements in the psychophysiological detection of concealed information.

It was on this background that Keeler developed the first instrument incorporating the essential features of the modern polygraph. Keeler also refined the relevant-irrelevant test, in which crime-relevant questions are interspersed with irrelevant questions. He experimented with personally embarrassing questions and surprise questions in an attempt to introduce a control stimulus that would evoke a reaction in innocent as well as guilty subjects.

Summers (1939) provided the first descriptions of the control question technique, in which a third type of question is employed that is designed to elicit an emotional response in both guilty and innocent subjects for the sake of comparison with the responses to relevant questions. Previously Larson (1921, p. 396) had referred to a technique in which "a control question, or one not concerning the subject under investigation, and yet calculated to stimulate various emotions, was alternated with one pertinent to the investigation." He did not, however, clearly distinguish between control and irrelevant questions.

Reid (1947; Reid & Inbau, 1977), often credited with first introducing the control question technique, did in fact refine this technique to essentially the same form in which it is practiced today. He also introduced the recording of muscular activity as a means to detect attempted countermeasures. Reid and his colleague, Fred Inbau (Reid & Inbau, 1977; Matte, 1980) did much to systematize the methods for questioning and data acquisition in interrogative polygraphy. Over the years they accumulated voluminous data on the applicability and effectiveness of various techniques of conventional interrogative polygraphy.

INTERROGATIVE POLYGRAPHY: THE CURRENT STATE OF THE ART

Types of Questions

The questions asked of the subject in %conventional interrogative polygraphy fall into four categories: "relevant questions," "irrelevant questions," "control questions," and "concealed information questions." Relevant questions are directly related to the focus of an investigation. Irrelevant questions are irrelevant to the investigation and are structured so as to have little or no emotional significance for the subject. The response to these questions provides a baseline: they establish the typical pattern with which the subject responds to routine questions. Control questions are questions that, while not directly relevant to the situation or issue under investigation, are designed to elicit an emotional response from the subject that is similar to the response that the relevant questions will elicit in a guilty subject. The control questions are concerned with various undesirable behaviors in which the subject may have been involved in the past, and are designed to elicit concern and/or doubt in the subject about the veracity of his response. These questions are also used for the purpose of comparison; they establish the typical physiological response of a subject to questions about which he is concerned. Concealed information questions focus on information about the crime or issue under investigation that would be known only to the guilty party. It is assumed that a guilty individual will respond more vigorously to (that is, be more physiologically aroused by) correct details associated with the crime than to incorrect details, whereas innocent subjects lacking in knowledge of the correct details will have the same response to both classes of items.

The Three Classes of Conventional Interrogative Polygraphy Tests

These four classes of questions are employed in three different interrogatory systems: 1) the relevant/irrelevant test, 2) the control question test, and 3) the concealed information test.

1) The Relevant/Irrelevant Test

In the relevant/irrelevant test (Larson, 1922, 1932; Keeler, 1930), two types of questions are presented: relevant and irrelevant. It is presumed that innocent people will respond in a similar manner to the questions from both classes, because they are unconcerned about the crime. Guilty persons will be more aroused by the relevant questions, because they are more concerned about the crime.

This technique has serious drawbacks, the most obvious of which is that the relevant questions concern subject matter that is inherently more upsetting than the irrelevant questions. Some innocent people, as well as guilty individuals, may respond to "Did you shoot John Jones last Tuesday?" or "Have you ever used cocaine?" more strongly than to "Do you live on Cherry Street?" Therefore a large response to the relevant questions may not be indicative of deception; rather it may be the result of the emotional tendencies and the physiological lability of the subject. Moreover, in the cases where the subject does not respond to the relevant questions, there is no indication how the subjects would have responded had the questions been about criminal activity of which the subject was guilty. In short, there is no control, and as a result both false negatives and false positives will occur. Because of this, the relevant/irrelevant technique is seldom used in investigations of a specific crime or situation, although it is still used in nonspecific investigations such as pre-employment and personnel screening.

2) The Control Question Test

The most frequently employed technique in the investigation of specific crimes or critical situations is the control question test, which was developed to solve some of the problems of the relevant/irrelevant test (Reid & Inbau, 1977; Kircher, Horowitz, & Raskin, 1988). In the control question test, control questions are added to relevant and irrelevant questions. These control questions are designed to elicit an almost obligatory emotional response. It is assumed that an innocent subject will respond more strongly to the control questions, since he/she should not be concerned about the crime or issue at hand. Guilty persons are expected to respond more to the relevant questions, since they are more concerned with the specific issue under investigation than with a general question about something undesirable they may have done in the past.

Summers (1939) referred to the control questions as "emotional standards." Their purpose was "to evoke within the individual rather intense psychogalvanic reactions due to surprise, anger, shame, or anxiety over situations which he would ordinarily prefer to conceal" (p. 341).

The usual control questions are somewhat vague, and cover a long period of time. They are designed to encourage the subject to lie, or at least to be unsure of the truthfulness of his or her answer (Reid & Inbau, 1977). Such control questions are sometimes referred to as "probable lie" questions. An alternative control question is the "directed lie" control question. These are questions regarding past activities, usually very minor negative activities that everyone is assumed to have committed, to which the subject is instructed to lie.

The control question test has been the subject of two classes of studies: 1) field studies involving actual crimes (e.g., Raskin, 1976; Horvath & Reid, 1971; Barland & Raskin 1976; Bersh, 1969; Davidson, 1979; Horvath, 1977; Hunter & Ash, 1973; Kleinmuntz & Szucko, 1984; Sicwick & Buckley, 1975; Wicklander & Hunter, 1975; Raskin, Barland, & Podlesny, 1978), and 2) studies of mock crimes or similar deception-oriented circumstances conducted in laboratories (e.g., Barland & Raskin 1975; Podlesny & Raskin 1978; Raskin & Hare 1978; Rovner, Raskin, & Kircher, 1978; Widacki & Horvath 1978; Dawson 1980; Hammond, 1980; Bradley & Janisse 1981; Szucko & Kleinmuntz, 1981; Ginton, Dale, Elaad, & Ben-Shakhar, 1982; Kircher & Raskin, 1982; Honts, Raskin, & Kircher, 1983; Kircher, 1983; Bradley & Ainsworth, 1984; Gatchel, Smith, & Kaplan, 1984; Honts, Hodes, & Raskin, 1985; Forman & McCauley, 1986).

The results of these studies have been the subject of considerable controversy over the effectiveness of the technique (see, for example, Lykken, 1978, 1979, 1988; Raskin, 1978, 1987, 1988; Raskin & Podlesny, 1979). Accuracy rates for detecting mock guilty subjects have ranged from a low of 71% (Szucko & Kleinmuntz, 1981) to a high of 100% (Dawson, 1980; Dale, Elaad, & Ben-Shakhar, 1982; Raskin & Hare, 1978). Accuracy for the detection of mock innocent subjects has generally been found to be lower, ranging from 49% correct (Szucko & Kleinmuntz, 1981) to 97% (Kircher & Raskin, 1982). A number of factors may contribute to the differences in accuracy rates reported. Studies varied widely in the type of subjects, the motivation of the subjects (higher motivation typically yielded higher detection accuracy), access by examiners to information (e.g., behavioral data) not included in the charts, the amount and type of psychophysiological information provided to the interpreters, the skill of the interpreters, and the method of scoring the charts.

One major difficulty with mock crime and other laboratory studies is that it would be exceedingly difficult (not to mention highly unethical) to subject individuals participating in a study to anything approaching the level of negative emotional arousal experienced by individuals who are actually guilty of major crimes in examinations attempting to detect that fact.

A few laboratory analog studies have provided a somewhat realistic experience. Ginton, Dale, Elaad, and Ben-Shakhar (1982) gave cadets at the Israeli Police Academy an opportunity to cheat on scoring a test they had taken, where the paper was chemically treated to reveal any changes. The officers, some of whom had cheated, were told that they were suspected of cheating, and offered an opportunity to take a polygraph test. They were told that their careers might depend on the outcome of the test. Such manipulations, though undoubtedly at least somewhat effective in simulating an actual field polygraph examination, are highly questionable on ethical grounds, and undoubtedly would not pass a review by a human subjects committee in the United States.

Accuracy rates for field studies have varied even more than those of mock crime studies. Accuracy of detection of criterion guilty individuals has ranged from 71% (Bersh, 1969) to 99% in one condition reported by Wicklander & Hunter (1975). Innocent subjects have been detected with accuracies ranging from 12% (with judicial outcome as the criterion of guilt, reported by Barland & Raskin, 1976) to 94% in one condition reported by Bersh (1969). In addition to many of the same variables affecting analog studies, field studies have varied in the method of selection of cases, the type of crimes, the suspected role in the crimes of the individuals examined, and the criterion for ground truth.

Ground truth is a particularly difficult issue, since in field studies it can never be known with certainty. Confession is held by some to be the most accurate criterion (Raskin, 1989), since it is rather unlikely that a suspect will confess who is not indeed guilty. However, confessions are not independent of the polygraph outcome, and tend to inflate the accuracy of both innocent and guilty determinations by systematically excluding cases where an error has been made. If a guilty subject is falsely found innocent, it is less likely that he or she will confess, and more likely that the investigators will continue mistakenly to seek a suspect who is guilty-which they will not find. Thus, no one will confess, the false negative will go undetected, and the case will be excluded from the analysis. If, on the other hand, an innocent suspect is falsely found guilty, and does not confess, the investigators may tend to lessen their efforts to find the guilty party since they think they already have found him. Again, no confession takes place, the false positive goes undetected, and the case is excluded from the analysis. (In the rare case in which an innocent subject does confess after failing the polygraph test, he is mistakenly scored as a correct detection.)

The other common methods of establishing guilt or innocence in field studies are judicial outcome and decision by a panel of experts. These methods, too, have their share of difficulties (Barland, 1982; Raskin, 1988). The judicial system in the United States, where virtually all of the field research has taken place, is based on the principal of "innocent until proven guilty beyond a reasonable doubt." Thus, both judges and juries and panels of experts who are trained in this legal tradition often tend to err on the side of finding the subject innocent. Also, in both judicial and panel decisions, the quantity of evidence available for consideration may be low, in which case the subject also will generally not be found guilty. This situation is a favorable one for human rights, of course, but it renders these methods of determination of culpability problematical for scientific purposes.

In an attempt to reduce some of the variability across examiners in the interpretation of polygraph charts and to supplement examiner interpretation of polygraph data, computerized statistical techniques have been developed which extract information from polygraph data (Kircher, 1983; Kircher & Raskin, 1981, 1988). The accuracy of this more objective approach is similar to the accuracy of examiner judgments.

The control question test is sometimes used in screening situations to investigate particular classes of activities or future intentions. This involves questioning similar to that employed in screening uses of the relevant/irrelevant technique, with the addition of control questions. Such a use of the control question test involves a difficult balance between control questions of a general nature designed to elicit an emotional response in both innocent and guilty subjects, and relevant questions, also often of a general nature, designed to elicit a response only in the guilty.

A few studies have been conducted on the use of polygraphy in actual government security screening applications. In a study by the Director of Central Intelligence (U.S. CIA, DCI, 1982) on background investigations in the CIA, the polygraph was used in an attempt to evaluate adverse information that had arisen regarding individuals being considered for employment or security clearances. In two-thirds of the cases where the information was resolved against the individual, the individual admitted the adverse information. There is no indication of whether the adverse information was correctly identified in the remaining one-third of the cases; and since the polygraph was used only after a thorough investigation had already revealed evidence against the individuals, it is impossible to know precisely what role the polygraph played or how effective it would have been in the absence of such an investigation and such pre-existing evidence. Edel & Jacoby (1975) showed that polygraph examiner ratings of a "physiological reaction" in applicants for U.S. government employment were reliable across examiners; but they did not determine whether the reaction actually was indicative of deception. In a laboratory study analogous to government security screening applications that used military intelligence personnel as subjects, Barland (1981) used the directed lie technique. Three different blind analysis techniques were used to identify subjects who were lying to any question, and the same three techniques were uses to identify which specific questions elicited lie responses. The average accuracy in identifying both truthful/lying subjects and truth/lie responses to specific questions was 69%.

3) Concealed Information Tests

Concealed information techniques are of two kinds: the guilty knowledge or concealed knowledge test and the peak of tension test. Both work on the premise that a guilty person will exhibit a larger response when confronted with correct details relevant to the crime under investigation than to similar details that are unrelated to the crime. An innocent person is expected to respond identically to both. The guilty knowledge test has been proposed as an alternative to control question techniques (Lykken, 1959, 1960, 1981, 1988; Giesen & Rollison, 1980; Ben-Shakhar, Bar-Hillel, & Lieblich, 1986; Furedy & Helsgrave, 1988). Lykken (1959) is often credited with the development of the guilty knowledge technique, although in fact it was described as early as 1908 by Munsterberg (1908), and its successful application was reported and analyzed in detail by Keeler (1930). In this technique, subjects are asked a variety of questions about details of the crime that would be known only to the guilty party. It is assumed that a guilty person will respond differentially to the correct, crime-relevant details. Scientific studies of the guilty knowledge test, like studies on the control question test, have produced mixed results (Lykken, 1959; Davidson, 1968; Podlesny & Raskin, 1978; Balloun & Holmes 1979; Giesen & Rollison, 1980; Bradley & Janisse, 1981; Stern, Breen, Watanable, & Perry, 1981; Bradley and Warfield, 1984; Iacono, Boisvenu, & Fleming, 1984; Elaad & Ben-Shakhar, 1989). It has been studied almost exclusively in the laboratory. Because the innocent subject does not know which are the relevant questions, the guilty knowledge test has often achieved a very low level of false positives (e.g., 0% in Lykken, 1959; Davidson, 1968; and Podlesny & Raskin, 1978). False negatives, however, have been a significant problem, sometimes reaching levels of about 40% (e.g., Balloun & Holmes, 1979; Bradley & Janisse, 1981).

In the peak of tension test (Harrelson, 1964), a question concerning a relevant and correct detail about a crime is embedded in a series of questions mentioning similar but irrelevant details. The sequence of presentation of the questions is known in advance to the subject. It is assumed that the physiological response will peak at the time of the relevant question (Barland & Raskin, 1973; Lykken, 1981).

Theory in Conventional Interrogative Polygraphy

In the two thousand years since Erasistratos made his successful discoveries, a large and rich data base has accumulated regarding the appearance and measurement of the changes . . . indicative of the inward passions and inclinations of the soul" (Plutarch, first century A.D./1952). Much more is now known about the mechanism through which the autonomic nervous system, when activated by emotion, brings about the changes measured by polygraphers. (See, for example, Lacey, 1967; Obrist, Webb, Sutterer, & Howard, 1970; Porges & Coles, 1976; Schwartz, Weinberger, & Singer, 1981; Kandel and Schwartz, 1985; Coles, Donchin, & Porges, 1986; Mathews, May, Mogg, & Eysenck, 1990; Davidson, Ekman, Saron, Senulis, & Friesen, 1990; Ekman, Davidson, & Friesen, 1990; Lang, Bradley, & Cuthbert, 1990; Eysenck, Mogg, May, Richards, & Mathews, 1991).

The fundamental psychophysiological theoretical foundation driving modern conventional interrogative polygraphy, however, is fundamentally the same as that which led Erasistratos to his early success in the psychophysiological detection of concealed information. One addition of modern interrogative polygraphy is the understanding of the adaptive value that these psychophysiological changes have in a physically dangerous situation.

Emotions consist of intensified feelings regarding a situation, which are caused by an interaction between the mind and the body resulting in physiological changes within the body to cope with the situation . . . When these feelings are strong . . . the accompanying physiological changes are extensive. Under strong fear, these changes have the effect of preparing the body for a fight or flight . . .

The system that prepares the body's defenses to meet these emergencies is...the sympathetic subdivision of the autonomic nervous system . . . In a polygraph situation, it is fear, fear of detection, fear of the consequences if the individual is detected, that causes the sympathetic system to activate in order to prepare the body to meet the emergency (Matte, 1980).

This modern description of the phenomenon is very similar to the ancient descriptions advanced by Erasistratos, and also to the descriptions set forth by early interrogative polygraphers. Munsterberg's (1908) explanations have been quoted above. Benussi (1914) attributed the physiological changes he measured to "internal excitement."

Marston (1917, 1938) analyzed in some detail the effect of the emotions of fear and rage on the sympathetic nervous system, and reached similar conclusions (although some of the details of Marston's analysis have not found favor with later experts). Similarly, Burtt (1921a, 1921b), after extensive experimentation and debriefing of subjects, implicated "fear or excitement." Keeler (1930) held that fear was the primary emotion behind the responses of a guilty subject.

There has been considerable controversy in recent years over the emotional/physiological arousal paradigm in interrogative polygraphy. One major contributing factor to this controversy may be the gulf that has developed between practicing polygraphers and academic scientists.

The ancient proponents of the emotional/physiological arousal paradigm for psychophysiological detection of concealed information were physician/scientists. Many of the early proponents of the introduction of modern instrumentation in the service of this paradigm in the first third of the twentieth century were scientists well within the mainstream of current scientific theory and methodology. They used the same equipment in interrogative polygraphy that they used for other, unrelated physiological and psychophysiological research (Munsterberg, 1908). After the introduction of the "polygraph" as a specific piece of machinery with the sole purpose of lie detection, however, a new breed of experts arose. The original field of expertise of these individuals was not physiology or psychophysiology but criminology and law enforcement.

Their expertise was primarily developed not in the scientific laboratory but in the applied, field setting, where both the theoretical and practical considerations held to be important differed from those that caught the attention of the academic psychophysiologists of the day. Critics and proponents alike agree that there was little contact between practicing polygraphers and the academic community. As Raskin (1979) stated, "It is interesting to note that field polygraphers (Backster, 1962; Reid, 1947) have managed to develop effective procedures for detection of deception in the field setting without the benefit of formal training in psychology and physiology and with minimal contact with the academic--scientific community."

Outside of the field of interrogative polygraphy, considerable attention was given to distinguishing between the psychophysiological manifestations of different emotions and developing a more detailed account of the psychophysiological signature of different emotional responses (Ax, 1953; Vanderhoof & Clancy, 1962; Lacey, 1967; Lacey & Lacey, 1970; Obrist, Webb, Sutterer, & Howard, 1970; Porges & Coles, 1976; Schwartz, Weinberger, & Singer, 1981; Mathews, May, Mogg, & Eysenck, 1990; Davidson, Ekman, Saron, Senulis, & Friesen, 1990; Ekman, Davidson, & Friesen, 1990; Lang, Bradley, & Cuthbert, 1990; Eysenck, Mogg, May, Richards, & Mathews, 1991). It has been shown, for example, that a unified theory of arousal is inadequate to explain some observed phenomena. Psychophysiologists have distinguished between different responses such as the orienting reflex and the defensive reflex: skin conductance increases in both cases, whereas heart rate decreases with the orienting reflex and increases with the defensive reflex.

Such distinctions, however, have by and large not been important for the actual practice of conventional interrogative polygraphy (Reid & Inbau, 1977).

Established on the original theoretical foundation recognizing the link between emotions and physiological responses, interrogative polygraphy has proceeded primarily empirically (Department of Defense, 1984; Raskin, 1979).

A strong psychophysiological response of any one of a number of flavors has been deemed sufficient to reveal that a relevant stimulus is emotionally arousing for an individual. In practice, polygraphers have come to recognize a wide variety of different, sometimes opposite responses as indicators of the state or states of heightened emotion that are hypothesized to accompany deception (Reid & Inbau, 1977).

A person's body functions at one physiological pace during chart time...whether fast, slow, calm, "nervous," or excited. This pace will henceforth be referred to as "norm." The instrument merely records this norm and any deviation therefrom. A deviation from norm, commonly called a response or reaction, is the result of a verbal stimulus (a question) . . . If a certain brain center interprets the question to mean harm to a person's well-being, providing the question is answered truthfully, a series of nerve impulses is generated. Specifically, the brain has delegated responsibility for protection of the body to the autonomic nervous system and its emergency subdivisions, the sympathetic and parasympathetic. As this emergency system goes into action, a minor to major change is recorded on a moving chart. (Ferguson & Miller, 1973, p. 146).

Although polygraphers have described in considerable detail the neurological mechanisms through which emotions and the concomitant autonomic nervous system activities are revealed in polygraph recordings (Ferguson & Miller, 1973), and some have indicated the need for some degree of specificity in the emotions elicited during a polygraph test (Matte, 1980), a fine-grained analysis of different emotions and their concomitant psychophysiological responses has not been necessary for the practice of conventional interrogative polygraphy. Some go so far as to reject such analysis as irrelevant.

First and foremost, it must be emphatically stated that the polygraph examiner is not concerned with attempting to differentiate one emotion from another. He is only concerned that a verbal stimulus did provoke an emotion which produced a sufficient nerve impulse to create a deviation from norm on the chart (Ferguson & Miller, 1973, p. 168).

The primarily empirical approach of conventional interrogative polygraphy is exemplified by Reid & Inbau (1977) in their authoritative text that has become perhaps the greatest classic in the field. They describe in voluminous detail the various psychophysiological responses that have proven in their extensive experience and that of other experts to be indicative of deception. They present a comprehensive and detailed account of the procedures that have been shown to be most effective in producing these responses; considerable knowledge about the art and science of interrogation aside from psychophysiological techniques; numerous case histories; a thoughtful discussion of legal, philosophical and moral issues; historical perspective; and reviews of research. The psychophysiological theory regarding the connection between the mind, emotions, and body that underlies psychophysiological detection of concealed information is only very briefly mentioned.

Although polygraphy has in general been practiced with a minimum of theoretical elaboration, there have been from time to time some notable exceptions to the rule. A number of scientists, some of them expert practicing polygraphers as well, have attempted to elaborate the emotional/physiological arousal paradigm for lie detection. Some have attempted also to bridge the gap between contemporary academic science and interrogative polygraphy as practiced.

In a classic paper on the subject, Davis (1961) outlined three specific theories for the psychophysiological response measured by conventional interrogative polygraphy. He called them the conditioned response theory, the conflict theory, and the threat-of-punishment theory. According to the conditioned response theory, "the critical questions play the role of conditioned stimuli, and evoke some `emotional` response with which they have been associated in the past." One difficulty with the conditioning theory is that it is insufficient to explain psychophysiological responses to lying about rather trivial matters with which the subject has had no experience in the past, such as which number is on a card.

The conflict theory "would presume that a specially large physiologic disturbance would occur when two incompatible reaction tendencies are aroused at the same time." The assumption is that a subject has the habit of telling the truth, but in a situation necessitating deception to maintain his/her innocence the subject would simultaneously experience an opposite tendency.

The punishment or threat-of-punishment theory, which has more often been called the fear theory, is the working understanding on which conventional polygraphy is generally practiced (Reid & Inbau, 1977; Barland & Raskin, 1973). The fear theory holds that "a person will give a large physiologic response during lying because he anticipates serious consequences if he fails to deceive. In common language it might be that he fails to deceive the machine operator for the very reason that he fears he will fail. The `fear` would be the very reaction detected." (Davis, 1961). To explain successful detection where the consequences are rather trivial, for example in the card test described above, one must interpret punishment rather broadly. Davis holds that perhaps the subject is psychologically "punished" by the detection of even trivial information.

Barland & Raskin (1973) describe another theory that they call the arousal theory. According to the arousal theory, "detection occurs because of the different arousal value of the various stimuli."

These elaborations of the basic theoretical foundation for interrogative polygraphy are not mutually exclusive. Moreover, it is likely that several different aspects of the psychophysiological phenomenon underlying conventional lie detection take place simultaneously (Davis, 1961; Barland & Raskin, 1973; Raskin, 1989).

Raskin (1979) hypothesizes that the response explained by the fear theory is primarily a defensive reflex, and the response explained by the arousal theory is primarily an orienting reflex. Which response is elicited depends on the interrogative situation in situations involving a relatively high level of emotion--particularly the response to relevant questions in a real-life application of the control question technique--the fear theory and the accompanying defensive response predominate. In situations involving a comparatively low level of emotion, the arousal theory and the accompanying orienting response predominate. Such low-emotion situations include mock interrogations, and especially mock interrogations using the guilty knowledge test. (Note that virtually all of the published results on the guilty knowledge test have been obtained in mock, laboratory situations.) The response to control questions has also been explained in terms of the arousal theory (Raskin, 1979).

The above hypothesis is in accord with research that has recorded somewhat different physiological responses in different lie detection situations. The electrodermal response is particularly effective in detecting the orienting reflex, and may be unstable in situations of very high emotion (Reid & Inbau, 1966). Cardiovascular and pulmonary responses, by contrast, are more marked in situations involving very high emotional levels. This may explain why Reid & Inbau (1966) found the cardiovascular and pulmonary measurements to be superior to the electrodermal measurement for field use, but not in the laboratory. (Note, however, that in the 1977 revision of their book, Reid and Inbau modified their conclusions to include a somewhat more favorable evaluation of the electrodermal response.) Other researchers (e.g., Podlesny & Raskin, 1977), on the basis of scientific studies many of which were laboratory studies involving mock crimes, concluded that the electrodermal response was superior. The predominance of the arousal theory in low emotion situations, combined with the fact that the guilty knowledge test has been studied almost exclusively in mock situations and is inherently less stressful than the control question test, may similarly explain the finding (Podlesny & Raskin, 1978; Balloun & Holmes, 1979; Bradley & Janisse, 1981; Iacono, Cerri, Patrick, & Fleming, 1987) that the electrodermal response is the most, and perhaps the only, effective measurement in the guilty knowledge test.

Different researchers have emphasized a different balance between the fear and arousal responses. Marston (1917; 1932) attempted to exclude the arousal response and focus on measurements that would emphasize the fear response. Ben-Shakar, Lieblich, and Kugelmass (1975), in a low emotion laboratory study on the guilty knowledge technique, attempted to minimize the fear response, and to obtain discrimination of knowledge on the basis of the arousal theory. They formulated the dichotomization hypothesis, which attempted to explain their results as differential habituation of the electrodermal response to different categories of stimuli (relevant and irrelevant).

There is no evidence in the scientific literature, however, that a guilty individual accused of a serious crime will fail to display a fear response when confronted during a guilty knowledge test with relevant details of the crime. On the contrary, anecdotal field experience (Keeler, 1930, p. 49) indicates a "violent emotional response." Nor is there convincing evidence that the arousal response can (or should) be entirely excluded in an actual interrogation situation (see Raskin, 1979, 1987). Fortunately, the proponents of the various points of view within the conventional paradigm agree that for practical purposes it does not matter. As long as the subjects' responses are different in magnitude to the different types of questions, the direction and specific details of the response are not crucial for distinguishing guilt from innocence. As Lykken (1959), a major proponent of the guilty knowledge test and severe critic of the control question test, noted in describing the guilty knowledge test,

A guilty subject would be expected to respond differently to the relevant than the irrelevant items. Usually, he would be expected to give larger responses to the relevant items, although it should be pointed out that any consistent difference in the responses to the two classes of stimuli is evidence of guilt.

Note that in this regard Lykken's view is very similar to that expressed by Ferguson & Miller (1973), supporters of the control question test, as quoted above in reference to that technique.

Conventional interrogative polygraphy, and in particular the control question test, have been criticized on a number of scientific, ethical, humanitarian, and legal grounds (Skolnick, 1961; OTA, 1983; Saxe, Dougherty, & Cross, 1985; Kleinmuntz & Szucko, 1982; Iacono, 1985;). There has been what might be termed a lively interchange of ideas between proponents of the different conventional polygraphy methods (see, for example, Lykken, 1978; Raskin, 1978; Barland, 1985; Horvath, 1985). Many of the harshest critics of the control question test are advocates of the guilty knowledge test (see, for example, Lykken, 1959, 1978, 1985, 1988; Furedy & Helsgrave, 1988; Ben-Shakhar, Bar-Hillel, & Lieblich, 1986), and the reverse is also true (Raskin, 1989). Criticisms range from a general statement that more research is needed to comparisons of control question polygraphy with the superstitious bone-pointing procedures used by Australian aborigines (Furedy, 1987). The evidence on the accuracy and validity of the conventional polygraphy is mixed, with estimates of accuracy ranging from chance to 100%. Both sides of the controversy (see, for example, Lykken, 1988; Raskin, 1988) conclude that the methodologically and theoretically sound studies support their point of view, whereas the studies purporting to support the opposite point of view are seriously flawed (Barland, 1985). There is general agreement that the quality and experience of the examiner have a significant effect on the outcome of a polygraph examination (Barland, 1988), although whether this is a strength or a weakness of the technique is a matter of debate. There also has been considerable controversy over who is qualified to speak with authority on interrogative polygraphy--experienced field polygraphers or academic scientists.

The current state of the emotional/physiological arousal paradigm for lie detection, then, is one of contradictions and controversy. Even the strongest proponents of the current paradigm acknowledge that it has both theoretical and practical weaknesses, while even its most vehement critics at least admit that conventional polygraphy has some degree of effectiveness in eliciting confessions from guilty subjects. In the next section I will compare the several applications of this paradigm that are currently in use, discuss their strengths and weaknesses, and contrast the current paradigm with the new paradigm reported herein.

Comparative Strengths and Weaknesses of the Conventional Techniques

The Relevant/Irrelevant Test

The fundamental weakness of the relevant/irrelevant technique is that the dimension of "relevant/irrelevant to this particular crime" is entirely confounded with the dimension of "having to do with any crime or undesirable activity (and consequently potentially upsetting)." This can result in either false positives or false negatives (Podlesny & Raskin, 1977). If a subject is particularly responsive to any mention of criminal or undesirable activity, then a false positive can result. If an individual is particularly unresponsive to such mention, then a false negative can result. Even many of the strongest supporters of conventional polygraphy now agree that the relevant/irrelevant test is seriously flawed (Raskin, 1979).

The control question test attempts to deal with this confound by introducing a class of questions, the control questions, which have to do with crimes or undesirable activities on the part of the subject, but are not relevant to the particular crime under investigation.

The Control Question Test

This aspect of control is the primary strength of the control question technique. In introducing an additional type of question, however, the control question test also inevitably introduces an additional source of variability. The control questions are never fully comparable to the relevant questions. Ordinarily, control questions are of a general nature and cover a long period of time. (For example, "Have you ever stolen anything from anyplace where you worked?") Relevant questions are often very specific. (For example, "Did you steal S200 from Smith's wallet last Tuesday?") A false negative may result if an individual responds excessively to questions of the type exemplified by the control questions. This can be a result of the examiner's choice of control questions, the psychological or physiological makeup of the subject, deliberate countermeasures, or any one of a number of other factors. If, for any of the above reasons, an individual tends to respond only slightly to the kind of questions of which the control questions are examples, then a false positive may result. Similarly, exceptionally large or small responses to specific crime-related questions can result in false positives or negatives respectively. These difficulties can be ameliorated by skillful interrogation and choice of control questions, but they are intrinsic to the design of the control question test and can never be eliminated entirely.

It is generally agreed by critics and supporters alike (OTA, 1983; Barland, 1985) that the control question test is more effective at discovering guilty subjects than at clearing innocent ones. This may be due, at least in part, to the fact that the relevant questions are recognized as such by both innocent and guilty subjects, and the relevant questions may tend to be inherently more upsetting than the control questions regardless of the guilt or innocence of the subject.

Another major difficulty of the control question test is that it can, and in fact in order to be successful it must, subject an innocent subject to considerable stress and negative emotional arousal. In the case of a guilty subject in an actual criminal interrogation, field experience indicates that the relevant questions are highly stressful in both the control question test (Gugas, 1979) and the guilty knowledge test (Keeler, 1930). It may be argued, however, that this stress is a result of the subject's criminal activity, rather than the test per se, and that a similar level of stress would inevitably be incurred in any interrogation of a guilty subject. For an innocent subject, however, there are two additional sources of stress in the control question test. The relevant questions, particularly if the crime is serious, may be experienced as highly stressful. Moreover, the control questions are designed to elicit a similar response in all subjects to the response to relevant questions in a guilty subject. Since the responses to the relevant and control questions are compared to make the determination of guilt or innocence, an innocent subject, in order to pass the test, must exhibit larger responses to the control questions than to the relevant questions. This means that the control questions must be designed to produce an even larger response than questions that falsely accuse a subject of a real crime.

Munsterberg (1908), one of the earliest modern advocates of the emotional/physiological arousal paradigm for lie detection, summarized its limitations perhaps as eloquently as anyone since.

. . experiment gives us so far not sufficient hold for the discrimination of the guilty conscience and the emotional excitement of the innocent. The innocent man, especially the nervous man, may grow as much excited on the witness stand as the criminal when the victim and the means of the crime are mentioned; his fear that he may be condemned unjustly may influence his muscles, glands and blood vessels as strongly as if he were guilty. Experimental psychology cannot wish to imitate with its subtle methods the injustice of barbarous police methods. The real use of the emotion-method is therefore so far probably confined to those cases in which it is to be found out whether a suspected person knows anything about a certain place or man or thing.

The Guilty Knowledge Test

The greatest strength of the guilty knowledge test is that an innocent subject cannot distinguish between the relevant and irrelevant items. This avoids a major source of false positives that occur with the control question test: that an innocent subject is emotionally aroused by the inherently upsetting and clearly recognizable relevant questions. In a properly structured guilty knowledge test, only a guilty subject can recognize the relevant questions as such, and consequently an innocent subject is unlikely to respond differentially to the relevant items (Lykken, 1988). The fact that the innocent subject does not know which are the relevant questions, the absence of accusatory relevant questions, and the lack of intentionally disturbing control questions also make the guilty knowledge test less stressful for an innocent subject than the control question test.

The major weakness of the guilty knowledge test is its lack of control. If a subject does not respond to the relevant questions, there may be at least two different explanations: 1) the subject is innocent, and is not responding to the relevant stimuli because they are no different to him than the irrelevant stimuli, or 2) the subject is guilty, but for some physiological or psychological reason does not show a differential response to questions relevant to his guilt. The guilty knowledge test does not allow a distinction between these two possibilities. If the subject does not respond, there is no way of determining what it would have taken to make him respond. This feature makes the guilty knowledge test particularly susceptible to false negatives (Raskin, 1988). (Note that the ERP-based test reported here does not suffer from this weakness. The "target" stimuli provide a control response that is lacking in the conventional guilty knowledge test).

Another weakness of the guilty knowledge test is that it necessitates knowledge of the details of the crime on the part of the examiner. In practice, this means that considerably more time and effort must be invested by an interrogator in the investigation. In some cases the guilty knowledge test can not be applied at all. For example, the investigators may be unable to obtain sufficient information to structure a test based on guilty knowledge; or a subject may admit being present at the crime scene but maintain that he or she was a witness or that no crime occurred..Another difficulty is that special care must be taken not to provide the suspect with critical information during interrogation prior to the test.

Because of these difficulties, the conventional guilty knowledge test has not been widely implemented in actual investigations in the United States. In Japan and Israel, however, it is much more widely employed.

Both the strengths and the weaknesses of the peak of tension technique are similar to those of the guilty knowledge technique.

Of course, the skill of the examiner is an important factor in the effectiveness of all conventional interrogative polygraphy techniques. Only a highly skilled examiner can use any of these techniques effectively (Reid and Inbau, 1977; Raskin, 1988).Each of the conventional techniques, however, suffers from particular inherent difficulties. Moreover, all of them suffer from some common difficulties derived from their dependence on the interactions between emotions and the autonomic nervous system.

The Autonomic Nervous System in Interrogative Polygraphy

All previous approaches to interrogative polygraphy are based on the assumption that a deceptive action on the part of the subject will be accompanied by an affective reaction. (For reviews, see Furedy, 1986; OTA, 1983; Department of Defense, 1984; Kircher, Horowitz, & Raskin, 1988; Raskin, 1987, 1988, 1990; Lykken, 1988). This reaction, in turn, is associated with activity in the autonomic nervous system, an activity that is manifested by an ensemble of actions generated within a variety of effectors driven by the autonomic nervous system. Some of these actions can be detected by monitoring several biological systems whose activity is modulated by the autonomic nervous system.

The use of affective responses in the detection of deception suffers from a number of inherent difficulties. All conventional detection of deception techniques depend on 1) the examiner's ability to create a situation that will effectively elicit particular emotions in the subject--and in particular, different patterns or levels of emotional arousal in innocent and guilty subjects--in the course of an interrogation, 2) the subject's emotional responses, 3) the subject's physiological responses when these emotions are aroused, 4) accurate measurement of these physiological responses, and 5) the examiner's interpretation of these physiological measurements.

Difficulties with Autonomic Nervous System-Based Polygraphy

Even if it is assumed that autonomic nervous system responses can be accurately measured and that the examiner will bring to the interpretation of these measurements a high level of expertise, a number of difficulties inevitably arise in drawing inferences about the guilt or innocence of a subject based on the above sequence of events. A researcher or interrogator is never certain that even a carefully structured and effectively executed interrogation will differentially elicit the requisite level of emotional arousal at the appropriate times. Subjects may exhibit unpredictable, unusual, or unexpected emotional responses (or lack of responses), or may manipulate their emotions to avoid detection. Some types of individuals (e.g., psychopaths) may not exhibit the emotions usually found. Different individuals may exhibit similar emotions for different reasons. For example, two individuals may be equally fearful in response to a particular item, one because he committed a crime and is afraid of detection and another because he did not commit the crime and is afraid of being falsely convicted.

Even if the emotional responses are as predicted and desired, autonomic nervous system responses may not be. Subjects may control their physiological responses voluntarily or through drugs, may employ hidden techniques (e.g., biting the tongue) to bring about misleading responses, or may naturally have unusually large or small autonomic .nervous system responses or idiosyncratic patterns of electrodermal response, breathing, or cardiovascular activity. The difficulty in inferring emotions from their autonomic nervous system "correlates" is well known.

As has been reported in many studies of interrogative polygraphy, false positives are often the consequence of an affective response to a question, or to the presentation of an item, for reasons that have little or nothing to do with deception (Lykken, 1974, 1978). Conversely, false negatives often result from the absence, or the low amplitude, of an autonomic nervous system manifestation of an affect.

It is also the case that only a portion of the variance in the autonomic nervous system activity is driven by affective processes. The autonomic nervous system is charged with the regulation of the most vital of the body's functions, and changes in heart rate, blood pressure, respiration and sweat gland activity reflect a variety of non-affective demands by the system. These include, for example, varying demands for blood supply, variations in the need for oxygen, and modulations of body temperature. From the point of view of interrogative polygraphy the autonomic nervous system is quite a noisy system. The "signal," that is, the autonomic nervous system activity driven by affect, may be swamped or masked by the "noise," the autonomic nervous system activity that is related to the vegetative and energetic functions of the system.

These factors are, in part, responsible for the ability of individuals to modify autonomic nervous system at will. The susceptibility of the autonomic nervous system for control using biofeedback makes it even more vulnerable, especially for specially trained individuals, to voluntary control by the subject. Countermeasures may range from the voluntary generation of affect by cognitive control to the activation of bodily parts so as to create energy needs that will be reflected in autonomic nervous system responses at critical points in the polygraph examination.