Respiration Line Length

Donald J. Krapohl & Donnie W. Dutton

Abstract

Line length of the respiration tracing has been found in polygraphy to be one of the reliable indicators of the act of deceiving.  Despite its value, respiration line length (RLL) is a concept unfamiliar to most polygraph practitioners.  RLL characterizes diagnostic information as a single value, and it may be useful for comparing phasic respiration responses that have different patterns.  The present paper is intended as an introduction to respiration line length for those who may be less familiar with it.

Introduction 

It has long been known that the respiration waveform contains information that permits a reliable inference of deception (Benussi, 1914; Burtt, 1921.)  The respiration channel was one of the first to be included in field polygraphs in the 1920s (Larson, 1923), and has remained in all polygraphs used for the detection of deception since that time.  Today respiration is recorded at not one, but two sites on the chest.  Though there is not uniform agreement in the field regarding respiration’s value relative to the other two polygraph data channels (Matte & Reuss, 1992; Bell, Raskin, Honts & Kircher, 1999), all would agree that respiration does make an important contribution to polygraph decision accuracy.

There is considerable less agreement among practitioners, however, on how to evaluate the respiration tracing.  In most scoring systems there are many more individual deception criteria taught for respiration than the other two polygraph tracings, with only some overlap of criteria between the different scoring systems.  Swinford (1999) identifies 11 respiration patterns that could be used to numerically score respiration tracings.  Other systems contain in excess of 20 criteria.  With so many criteria, it begs the question: is everything that happens in respiration an indication of deception?

The large lists of deception criteria may be a consequence of how respiration is ordinarily analyzed.  For the entire history of the profession, field examiners have adopted a pattern-matching approach to the scoring of the respiration channel.  Examiners are trained to recognize the signatures of phasic respiratory responses, such as suppression, inhalation/exhalation (I/E) ratio changes, apnea, bradypnea, and other terms.  Though these patterns do correlate with the arousal associated with deceiving, field practice had no single concept that captured the shared diagnostic information that is manifested in these different visual forms.  In other words, these patterns ostensibly have little in common, hence the need to create long lists of stereotypical patterns, instead of seeking a single feature that shows up in several shapes. In contrast, simple amplitude increases in the electrodermal and cardiograph channels are highly diagnostic.  Amplitude is easy to interpret, and more or less of it is sufficient for assigning scores.  The employment for several separate criteria for the pneumograph suggests at least some inefficiency, and may give rise to arbitrary rules on how to score the patterns (see the special edition on chart interpretation, Polygraph 1999, 28(1).)

Pattern matching is not without some validity, and the present writers want to reassure field examiners that we do not advocate its abandonment.  Patterns are certainly easy to teach novices, and communicate with practitioners.  This approach does not perform as well, however, when a scorer tries to judge the relative intensity of two different patterns.  Pattern matching simply does not easily allow across-pattern assessment of reaction intensity.  As an example, there is scarcely a uniform opinion regarding what to do when a scorer is faced with a suppression on a relevant question and an equally significant change in the inhalation (I/E) ratio on an adjacent comparison question.  There is no single higher rule to apply regarding which pattern is more diagnostic of deception.  Well, almost no rule.  Research over the last 20 years indicates that something called respiration line length (RLL) could be a prevailing principle to apply across patterns, and even among similar patterns.  Let us begin with a little history.

The Search for Parsimony 

Dr. Howard Timm, as a doctoral student at Michigan State University, was the first to report the RLL concept.  While Timm was conducting his graduate research on polygraphy (1979), Dr. Frank Horvath suggested that he measure electrodermal activity with a handheld device designed for measuring distances on maps, an instrument called a planimeter.  One day, when a volunteer subject did not appear for his polygraph appointment, Timm used the time to experiment with the planimeter on other parts of the polygraph tracings collected from previous subjects (Timm, 2001).  Before long he discovered a relationship between line length and veracity: the shorter the line for a given period of time, the more likely that the examinee had given a deceptive answer.  Line length appeared to be the common thread among the different signature patterns of deception in the respiration channel.  Timm later published his findings in the Journal of Police Science in Administration in 1982.

Meanwhile, masters candidate John Kircher of the University of Utah was testing the feasibility of a computerized scoring system for polygraph data.  Because computers are much better at measuring things than they are at pattern recognition, Kircher was seeking a diagnostic measurement for phasic respiration responses instead of trying to teach the computer field scoring criteria.  In 1981, when Kircher was conducting his masters research, Timm had not yet published his RLL paper, and Kircher experimented with several objective measurements, all of which performed significantly poorer than manual scoring (Kircher, 1981).  In 1983, a year after the Timm paper was published, now-Ph.D. candidate John Kircher applied the RLL method, and he also found it to be very diagnostic (Kircher, 1983). RLL subsequently became the response measurement for the CPS algorithm, and is now either implicit or explicit in several polygraph scoring systems, including the Utah Scoring System (Bell, Raskin, Honts & Kircher, 1999), the Rank Order Scoring System (Honts & Driscoll, 1987), the Department of Defense Polygraph Institute Scoring System (DoDPI, 2001a, 2001b), and the Objective Scoring System (Krapohl & McManus, 1999). 

What RLL Represents 

RLL is nothing more than a measure of how much tracing activity occurs in a defined period of time.  To help conceptualize RLL, imagine that the tracings were made of string, instead of ink.  If one were to cut the string once at stimulus onset and again after 10 seconds had past, then stretched the string to remove the curves, the string would have a fixed length, something that could be measured with a ruler or other tool.  In the respiration channel, almost all of the most diagnostic phasic patterns cause that string to shorten: suppression, I/E ratio changes, apnea, and bradypnea.  As a rule, the shorter the RLL is, the more intense the reaction is.  There is one notable exception, the upward shift of the respiration tracing from its baseline.  This pattern may or may not show a shortening of the RLL.

Using RLL 

RLL is an important feature, but is most useful under circumstances when comparisons between respiration responses are difficult.  For example, if one category of question elicits a phasic response, and its comparative question does not, RLL does not reveal anything that is not already obvious.  Conversely, if two questions are being compared against one another, and both have phasic responses, RLL can help determine which has more intensity, thereby allowing a better assignment of scores. 

There are several options at the examiner’s disposal for determining RLL.  The easiest, and the most commonly used, is simply the visual assessment of the tracings.  Examiners can estimate the RLL by “eyeballing” the line. Though less precise, this method is sufficiently reliable for most field applications.  For better precision, a measuring device is needed.  One such device is the planimeter, previously discussed.  A scorer needs only trace the respiration waveform from stimulus onset to a subsequent timepoint.  Kircher used 10 seconds as the time window, while Timm used 15 seconds.  For field use, there may not be meaningful differences between these windows, but an examiner must use a constant window for all measurements of responses to be compared to one another.  To do otherwise would affect the measurements, and the value of the RLL.

A better option for RLL measurement is to have the computer polygraph do the measuring.  Some computer polygraph manufactures already have or are considering software with this capability.  The advantage of automated measurement of the RLL is the perfect reliability and high precision of the measurements.  There is also a substantial savings of time, and a reduction in errors recording the RLL values.

Computerized instrumentation also affords other possible representations of the RLL.   Non-traditional data displays are relatively easy to accomplish with software, and we hope polygraph manufacturers are investigating better ways of displaying the traditional tracings.  One possibility is to produce an RLL tracing, a single EDA-like line created by a scrolling moving average of RLL across a chart.  The occurrence of a shortening of the RLL would be represented by a drop in the RLL tracing.  The RLL tracing could supplement the traditional respiration recording, adding objective information for examiners to use in their interpretation of the respiration data.  There may be even better ways of displaying this information waiting for software developers to discover.

Summary 

The limits imposed by the pattern-matching approach to respiration analysis can often be exceeded by the use of RLL.  RLL provides a metric for comparing respiration responses that are manifested in different shapes.  Moreover, with RLL, a degree of relative intensity of responses can be determined.  RLL is not intended to replace the pattern-matching method, since the latter is highly effective most of the time, and is also important for the identification of certain types of countermeasures.  However, RLL can be of added benefit when it is otherwise difficult to gauge differences in response intensities.  Examiners are encouraged to experiment with this concept, to confirm for themselves that RLL is, indeed, a reliable diagnostic feature.

References

Bell, B.G., Raskin, D.C., Honts, C.R., Kircher, J.C. (1999). The Utah Numerical Scoring System. Polygraph, 28(1), 1-9.

Benussi, V. (1914).  Die atmungssymptome der lüge [The respiratory symptoms of lying].  Archiv fuer die Gesamte Psychologie, 31, 244-273. (In German).  English translation printed in Polygraph, 4 (1), 52-76.

Burtt, H.E. (1921).  The inspiration/expiration ratio during truth and falsehood. Journal of Experimental Psychology, 4(1), 1-23.

Department of Defense Polygraph Institute (2001a).  Test Data Analysis (Manual Scoring Process) 7-Position Numerical Analysis Scale.  Ft. Jackson, South Carolina.

Department of Defense Polygraph Institute (2001b).  Test Data Analysis (Manual Scoring Process) 3-Position Numerical Analysis Scale.  Ft, Jackson, South Carolina.

Honts, C.R., & Driscoll, L.N. (1987).  An evaluation of the reliability and validity of rank order and standard numerical scoring of polygraph charts.  Polygraph, 16(4), 241-257.

Kircher, J.C. (1981).  Computerized chart evaluation in the detection of deception.  Unpublished masters thesis.

Kircher, J.C. (1983).  Computerized decision making and patterns of activation in the detection of deception.  Doctoral disssertation, University of Utah, Salt Lake City.  Dissertation Abstracts International.  University Microfilm # AAD 83-13152.

Krapohl, D.J., & McManus, B. (1999) An objective method for manually scoring polygraph data.  Polygraph 28(3), 209-222.

Larson, J.A. (1923).  The cardio-pneumo-psychogram and its use in the study of emotions, with practical applications.  Journal of Experimental Psychology, 5(5), 323-328.

Matte, J.A., & Reuse, R.M. (1992).  A study of the relative effectiveness of physiological data in field polygraph examinations.  Polygraph 21(1), 1-22.

Timm, H.W. (1979). The effect of placebos and feedback on the detection of deception. Dissertation Abstracts International, 40(12-B). University Microfilm No. AAD80-13808.

Timm, H.W. (1982). Analyzing deception from respiration patterns. Journal of Police Science and Administration, 10(1), 47-51.

Timm, H.W. (2001, Jan. 2).  Personal communication with first author.

This paper first appeared in Polygraph (2001) 30(1).  It is reprinted here courtesy of the authors and the American Polygraph Association.  The first author can be contacted via e-mail at dkrapohl@aol.com.