Distinctions in Behavioral Impulsivity: Implications for Substance Abuse Research

Donald M. Dougherty, Ph.D.1, Charles W. Mathias, Ph.D.1, Dawn M. Marsh-Richard, Ph.D.1, R. Michael Furr, Ph.D.2, Sylvain O. Nouvion, Ph.D.1, and Michael A. Dawes,
M.D.1 1University of Texas-Health Science Center at San Antonio—Department of Psychiatry 7703 Floyd Curl Drive, Mail Code 7792 San Antonio, TX 78229-3900 2Wake Forest
University—Department of Psychology 415 Greene Hall Winston-Salem, NC 27109
Abstract Objectives—Researchers have clearly implicated impulsivity as having a key role in substance use disorders, and comparisons of self-report measures suggest
there are measurably different components of impulsive behavior. However comparatively little research has been devoted to understanding the multidimensional nature of
this construct using laboratory measures of impulsivity that may be more sensitive to tracking changes across time. Many studies have measured impulsivity, but this
construct has been measured using methodologically different types of laboratory impulsivity paradigms that are often used in isolation. As a result, it is important
to determine whether some of the most frequently used types of behavioral measures of impulsivity account for unique variance. Methods—Here, we used factor analytical
techniques in two studies to evaluate the independence of three of the most commonly used behavioral impulsivity paradigms. First, a factor analysis was conducted
using previously collected data (n = 204), and second, data was gathered specifically to replicate and extend the results of our original analysis (n = 198). Results—
Both studies revealed three distinct factors, confirming our hypothesis of at least three components of impulsive behavior that can be measured by these methodological
approaches. Conclusions—These findings suggest that researchers should carefully consider their selection of laboratory-behavioral impulsivity measures, and that the
measure(s) selected should be related to the particular underlying processes relevant to substance use disorders and treatment success.
Keywords impulsivity; behavior; factor analysis; human; substance abuse
Introduction Several broad domains of evidence support the conclusion that there are important relationships between impulsivity and substance abuse.1 First, impulsive
groups have higher rates of substance abuse2–3 and psychiatric disorders that feature impulsive characteristics have some of the highest prevalence of co-morbidity
with substance use disorders.4–5 Second, substance abusers are more impulsive, and this effect is reported across users of many major drug classes, including:
alcohol6, nicotine7, heroin8, cocaine9 and ecstasy.10–11 Third, impulsivity has an
Corresponding Author: Donald M. Dougherty, Ph.D. University of Texas-Health Science Center at San Antonio—Department of Psychiatry 7703 Floyd Curl Drive, Mail Code
7793 San Antonio, TX 78229-3900 Phone: 210 562 6600 Fax: 210 562 6605 doughertyd@uthscsa.edu.
NIH Public Access Author Manuscript Addict Disord Their Treat. Author manuscript; available in PMC 2010 June 8.
Published in final edited form as: Addict Disord Their Treat. 2009 June 1; 8(2): 61–73. doi:10.1097/ADT.0b013e318172e488.
NIH-PA Author ManuscriptNIH-PA Author ManuscriptNIH-PA Author Manuscript
impact on substance abuse treatment. For example, among cocaine-dependent men and women entering a 12-week program for drug treatment, there was an inverse
relationship between impulsivity and retention in treatment.12 Given these relationships, impulsivity has been an area of intense research in the substance abuse
literature13, although questions remain as to how performance differences across impulsivity measures should be interpreted.
Studies using self-report and/or laboratory measures have shown that these tools appear to measure multiple aspects of impulsive behavior and that measures of
retrospective selfreported impulsivity tend to assess different aspects compared to laboratory-measured impulsivity.14–17 These findings illustrate the importance of
using multiple modes of measurement concurrently to assess this complex construct accurately, and the importance of determining the independence among different
measures to test different components of impulsive behavior. When examining the construct of impulsivity using self-report instruments, researchers have identified
from two to four different components, depending on the measures used and the samples tested. In an exploratory factor analysis of eighteen measures of impulsivity
administered to 437 undergraduate students, Whiteside and Lynam18 identified four independent components that they termed lack of premeditation, urgency, sensation
seeking, and lack of perseverance. Zelenski and Larsen19 conducted a principal factor analysis using five self-reports from 86 undergraduate student volunteers that
identified three components, which they termed impulsivity-thrill seeking, reward sensitivity, and punishment sensitivity. In another analysis of four impulsivity
questionnaire measures administered to 245 adults, Miller and colleagues20 also identified three components that they termed non-planning and dysfunctional impulsive
behavior, functional venturesomeness, and reward responsiveness and drive. Although, in a sample of 32 adult males (10 formerly substance dependent) Lane and
colleagues15 used three of the same self-report measures, but found that all measures loaded onto a single component. While these and other studies13, 21–22 provide
somewhat mixed results that are most likely due to methodological differences, the similarities suggest that measurably distinct components can be assessed with self-
reports.
Similarly, there have been studies that have examined the factor structure of impulsivity as measured by laboratory-behavioral tasks. For instance, the Lane and
colleagues’15 study also included behavioral measures. In a separate principal components analysis using just the behavioral tasks, the authors found impulsive
responding on a continuous performance test and delay-discounting task loaded together on a single factor, while an investigation by Reynolds and colleagues17, showed
that delay-discounting performance was a factor distinct from both CPT and stop-task performance. The different results of these studies could be accounted for by
differences in: (1) the samples and the sample sizes – Lane and colleagues sampled a relatively small group of 32 adult men, 10 of which were formerly substance
dependent, while Reynolds and colleagues sampled 99 young-adult college students (substance use disorders excluded); and (2) the number and the type of methodological
tasks used in the analyses – Lane’s included a differential reinforcement of low-rate behavior task and a delaydiscounting task, while Reynolds also included a risk-
taking task (i.e., Balloon Analogue Risk Task) that may be a measure of a separate construct. Thus, although there are likely to be multiple distinct factors of
laboratory-measured impulsive behavior, work in this area has been limited by the methodological restraints outline above, and further questions remain as to the
distinctiveness of patterns of behavior based on specific measures of impulsivity.
Differing results of the factor analyses of both questionnaire and laboratory measures highlight the fact that to examine a complex construct such as impulsivity
adequately, it is necessary to have a clear operational definition. One proposed definition, “a predisposition toward rapid, unplanned reactions to internal or
external stimuli without regard to the negative consequences of these reactions to the impulsive individual or to others”23, encompasses many aspects of impulsive
behavior. Based on this definition, impulsive behavior has been conceptualized as
Dougherty et al. Page 2
Addict Disord Their Treat. Author manuscript; available in PMC 2010 June 8.
NIH-PA Author ManuscriptNIH-PA Author ManuscriptNIH-PA Author Manuscript
involving at least three component processes.24 This three-component model (Figure 1) includes: (1) response initiation; (2) response inhibition; and (3) consequence
sensitivity. Response initiation can be defined as impulsive responding that occurs prior to complete processing and evaluation of a stimulus. Response inhibition can
be defined as a failure to inhibit an already initiated response. Consequence sensitivity (also conceptualized as reward sensitivity13, 21–22) can be defined as
responses that persist despite negative or less than optimal consequences (e.g., smaller reward or punishment).
The operational definitions of these three components encompass a number of different laboratory-impulsivity measures reported in the literature. There are several
types of continuous performance tests (CPTs25) that assess the degree to which participants respond selectively to a series of rapidly presented stimuli (e.g.,
letters, numbers, or abstract shapes). This is a type of response initiation measure where responses to stimuli other than designated targets have been interpreted as
impulsive because they represent anticipatory and incomplete stimulus processing that leads to rapid, but incorrect, responding.26–30 Different types of stoptasks
assess the ability to respond appropriately to target stimuli, but withhold responses when the target is accompanied by a “stop” signal. This type of task represents
the response inhibition component where responding to “stop” signal stimuli are interpreted as impulsive because they reflect an inability to inhibit an already
initiated response.31 There are also several types of delay-discounting tasks that assess the preference of a smaller-sooner reinforcement over a larger-later
reinforcement. The preference for a more immediate reinforcement despite a lower net gain is interpreted as the impulsive choice in the delay-discounting procedure.
While operational definitions applied to these laboratory measures of impulsive behavior are distinct, the extent to which these definitions are measurably distinct is
debatable.15 As stated by Reynolds and colleagues17, “further investigation of the role of impulsivity in substance abuse requires a better understanding of the
equivalence of different assessment methods and of the factor structure of impulsive behavior” (p. 306).
The purpose of the two studies presented here was to extend the results of previous studies that have examined differences among impulsivity measures. Using two large
samples of healthy individuals, we tested the independence of performance on three major methodological approaches to assess laboratory-behavioral impulsivity:
continuous performance, stop-signal, and delay-discounting tasks. These different methodologies are conceptually linked to the proposed model of impulsivity described
above.24 Our goal was to establish the factor structure of laboratory-measured impulsivity in normal healthy controls, which is the first step toward determining the
component, or profile of components, of impulsivity that may be predictive of the initiation, continued use, and/or treatment success in particular groups of substance
abusers.
In the first study, a factor analysis was conducted using previously collected behavioral performance data from healthy adolescents and adults recruited from a large
urban area. In the second study, data were gathered prospectively to replicate and extend the results of the first study by recruiting healthy adults. Participants
completed three computerized paradigms designed to measure different aspects of behavioral impulsivity: a continuous performance test, a stop task, and a delay-
discounting task. Factor analyses were conducted testing whether eight primary dependent measures of behavioral impulsivity derived from these laboratory behavioral
tasks would yield distinct factors. Given the theoretical distinctions between the meaning of performance on these tasks, we hypothesized that these eight dependent
variables derived from the above behavioral tasks would yield three correspondingly distinct factors.
Dougherty et al. Page 3
Addict Disord Their Treat. Author manuscript; available in PMC 2010 June 8.
NIH-PA Author ManuscriptNIH-PA Author ManuscriptNIH-PA Author Manuscript
Study 1
This study was an analysis of previously collected laboratory behavioral performance data of three measures of impulsivity among healthy control participants. These
data were obtained from previous research.32–37
Materials and Methods Participants—Two hundred four healthy adolescent and adult controls completed one session of each of the three laboratory behavioral measures
(described in the LaboratoryBehavioral Measures of Impulsivity section below). These participants responded to newspaper advertisements for participation in behavioral
research at the University of Texas Health Science Center at Houston. All participants completed a brief telephone screening followed by an on-site interview to
determine eligibility. This interview included a semi-structured psychiatric diagnostic interview, a health history, and a physical examination. The Structured
Clinical Interview for the DSM-IV38 was used to screen for psychiatric disorders in adults (i.e., 18 years and older), while the Schedule for Affective Disorders and
Schizophrenia for School-Age Children-Present and Lifetime Version (K-SADS-PL39) or the Diagnostic Interview Schedule for Children Version IV (DISC-IV40) was used to
screen for psychiatric disorders in adolescents (i.e., younger than 18 years). A subset of participants (n = 73 adolescents) also completed intelligence screening
(Wechsler Abbreviated Scale of Intelligence41). Based on this interview, participants were included if they: (a) had no Axis-I psychiatric disorder; (b) had no past or
present medical condition that would interfere with study participation; (c) had an IQ ≥ 80; (d) had a negative drug/alcohol test; (e) did not report use of
psychoactive medications; (f) did not smoke more than one pack of cigarettes per day; and (g) were not pregnant. Prior to participation, informed assent was obtained
from adolescents and consent from adults. This study and its procedures were reviewed and approved by the institutional review board of The University of Texas Health
Science Center at Houston.
Procedure—Once enrolled, participants completed the laboratory behavior testing between the hours of 0800 and 1630. Participants provided expired-air and urine samples
to screen for current alcohol use (Intoximeter 3000-III, St. Louis, MO) and to confirm the absence of major drugs of abuse (i.e., marijuana, cocaine, amphetamine, and
benzodiazepines; Syva Corp. Palo Alto, CA). Urine samples were also used for pregnancy screening for female participants (e.p.t.® Pregnancy Test; Pfizer, Inc., Morris
Plains, NJ, USA). Three behavioral measures of impulsivity were completed in a semi-counterbalanced order (i.e., because of task demands, the CPT and Stop-task were
never presented consecutively), and standardized instructions were given prior to each of the tasks. Tasks were completed in a sound-attenuated chamber equipped with a
computer monitor and two-button mouse. Participants earned points during each task that were exchanged for money, and earnings typically ranged between $6 and $8 per
task.
Laboratory-Behavioral Measures of Impulsivity Immediate and Delayed Memory Task (IMT/DMT): The IMT/DMT32, 42 is a modified Continuous Performance Test (CPT25) that has
been validated and shown to be sensitive to trait and state aspects of impulsive behavior in both adolescents and adults.33, 36–37, 43–46 In the IMT, participants were
required to respond selectively to a series of randomly generated 5-digit numbers (e.g., 38391) that appeared on the computer monitor for 500 msec at a rate of one per
second. The participants were instructed to respond when two identical numbers were presented in sequence. The DMT is similar to the IMT, but includes a distracter
stimulus (the number 12345, which is to be ignored) that appears three times between each of the numbers to be compared and responded to by participants. These
distracter stimuli were presented for 500 msec at a rate of one per second, which resulted in a 3.5 sec separation between the stimuli to be compared.
Dougherty et al. Page 4
Addict Disord Their Treat. Author manuscript; available in PMC 2010 June 8.
NIH-PA Author ManuscriptNIH-PA Author ManuscriptNIH-PA Author Manuscript
The primary types of stimuli are target, catch, and filler: (1) a target stimulus is a 5-digit number that is identical to the preceding number. Responses to target
stimuli were recorded as correct detections; (2) a catch stimulus is a 5-digit number that differs from the preceding number by only one digit (the position and value
were determined randomly). Responses to catch stimuli were recorded as commission errors; and (3) a filler stimulus is a random 5-digit number that appears whenever a
target or catch trial is not scheduled to appear. Responses to filler stimuli were recorded as filler errors. The presentation probability was 33% for either target or
catch stimuli and 34% for filler stimuli. The IMT and DMT each consisted of two 5-min testing blocks. The testing blocks alternated (i.e., IMT/DMT/IMT/DMT) and were
separated by a 30sec rest period.
The primary dependent measure for both IMT and DMT was the ratio of commission errors to correct detections. Commission errors are thought to result from anticipatory
or incomplete processing of the stimulus leading to a rapid, but incorrect, response44, and the ratio of commission errors to correct detections is used to account for
inter-individual performance differences in the ability to respond to target stimuli.43 Commission errors, and their proportion to correct detections, have been found
to be elevated in impulsive populations such as adults with an Axis I or Axis II diagnosis47, women with a history of physical fighting37, and boys with learning-
disabilities48, as well as following acute alcohol administration.43 GoStop Impulsivity Paradigm (GoStop): The GoStop49–50 is a stop task used to measure the failure
to inhibit responding when a targeted “go” cue is unpredictably coupled with a “stop” cue. In this task, 5-digit numbers were presented in rapid sequence. During a 15
-min session, random 5-digit numbers appeared for 500 msec at a rate of one every two seconds (i.e., 500 msec on, 1500 msec off). Half were target trials (matching
stimuli) and half were filler trials (novel stimuli). This task features “stop” trials for half of all target trials, where the matching number, presented in black
(the “go” cue), changed to red (the “stop” cue) at intervals ranging from 50 to 350 msec after the stimulus appeared. Participants were instructed to respond while the
matching number was still on the monitor, but not to respond if the number turned red. In other words, for half of the target (i.e., matching number) trials a “go”
stimulus (the matching number) included a “stop” cue where the number-color changed at varying delays. There were a total of 416 trials: 104 target, 104 stop, and 208
filler trials. For stop trials, there was an equal probability of one of four delays: 50, 150, 250, and 350 msec.
The primary dependent measures for the GoStop were the ratios of inhibition failures to the total number of stop trials for each stop delay. This ratio was calculated
separately for each of the four stop delays by dividing the number of responses made to stop trials (failures to inhibit a response) by the total number of go trial
responses (correct detections). This proportion yields information about the inability to stop an already initiated response, and has been found to be elevated among
impulsive populations.44, 51 Two Choice Impulsivity Paradigm (TCIP): The TCIP50, 52 is a discrete-choice delaydiscounting procedure that measures a participant’s
preference for smaller-sooner versus larger-later rewards. Variations of this type of paradigm are the most commonly used in human and nonhuman impulsivity studies.53
In the TCIP, participants made 50 reward choices by clicking on a circle or a square to add points to a counter. Following standardized instructions, participants
completed a brief practice session of individual presentations of five circle choices followed by five square choices. This practice session allowed the participant to
make associations between the two delay-reward contingencies without explicit information being provided during instructions. Following the practice session, circles
and squares appeared together on the monitor in black against a white background for each trial. The left/right orientation of the two shapes was randomly determined.
The participant chose between clicking on a circle to earn 5 points after waiting 5 sec and clicking on a square to earn 15 points after
Dougherty et al. Page 5
Addict Disord Their Treat. Author manuscript; available in PMC 2010 June 8.
NIH-PA Author ManuscriptNIH-PA Author ManuscriptNIH-PA Author Manuscript
waiting 15 sec (this delay-reward contingency was fixed throughout the 50 trials of this procedure). After choosing a shape, the other shape disappeared and the
selected shape faded to gray. After the scheduled delay elapsed, the shape changed back to black and flashed for 500 msec once per second; at this time the participant
clicked on the shape again to add the delay-contingent reward to the counter.
The two primary dependent variables were the proportion of smaller-sooner reward choices and the total number of consecutive larger-later reward choices.50, 52 The
proportion of smaller-sooner reward choices is an indicator of preference for the more impulsive choice. The total number of consecutive larger-later reward choices is
an additional indicator of tolerance for delayed reward.
Data Analyses—Performance on the IMT/DMT, GoStop, and TCIP behavioral impulsivity measures was analyzed to determine whether the dependent variables obtained from
these tasks assess independent aspects of impulsivity. There were two primary analytical strategies for making these comparisons. First, Pearson product-moment
correlation coefficients were computed to examine the relationship between each of the dependent measures of the impulsivity tasks. Second, a factor analysis was
conducted to examine the structural associations among the eight dependent variables: IMT and DMT ratios, GoStop 50, 150, 250, and 350 msec stop delay ratios, and TCIP
proportion of smaller-sooner and consecutive longerlater choices. Since correlations existed between some of the dependent measures of the different task types,
oblique (Promax) rotations were performed with Kaiser Normalization. Data were analyzed using SPSS® version 15.0 (SPSS Inc; Chicago, IL).
Results
Participant Characteristics—Forty male and 164 female (N = 204) participants completed the three behavioral impulsivity measures and were included in all analyses. Of
these, 131 adults (n = 8 men and 123 women) were from 18 through 52 years old (M = 31.8, SD = 9.0) with an average education of 13.6 years (SD = 2.2). The 73
adolescents (n = 32 boys and 41 girls) were from 13 through 17 years old (M = 14.6, SD = 1.3) with an average education of 9.2 years (SD = 1.6). All of the adolescents
were tested with the Wechsler Abbreviate Scale of Intelligence and had an average Full Scale Score of 101.6 (SD = 12.2). The total sample was ethnically diverse (45%
African American, 31% Caucasian, 18% Hispanic, 2% Asian, and 4% other/multi-ethnic), and of low to middle socioeconomic status (determined by the Four Factor Index of
Social Status: M = 39.0, SD = 11.3, n = 104).54
Behavioral Task Performance—Means of the IMT and DMT ratios (commission errors/ correct detections) were .31 (SD = .16) and .29 (SD = .19), respectively. For the
GoStop, ratios averaged .07 (SD = .13), .27 (SD = .25), .54 (SD = .28) and .72 (SD = .27) for the 50, 150, 250 and 350 msec stop-delays, respectively. For the TCIP,
the average of the proportion of short choices was .27 (SD = .23), and the average number of consecutive long choices was 19.7 (SD = 13.8).
Correlations: Pearson product-moment correlations were conducted among the eight dependent variables to determine the extent to which the three behavioral measures
were related (see Table 1). The highest correlations found were between dependent variables within each task type. For example, the IMT ratio correlated highly with
the DMT ratio. Similarly, the GoStop ratios were highly correlated with each other, as were the TCIP proportion short choices and most consecutive long choices. There
were smaller, but statistically significant, correlations between IMT/DMT ratios and several of the GoStop ratios.
Dougherty et al. Page 6
Addict Disord Their Treat. Author manuscript; available in PMC 2010 June 8.
NIH-PA Author ManuscriptNIH-PA Author ManuscriptNIH-PA Author Manuscript
Factor Analysis: A Principal Axis Factor Analysis with a promax rotation revealed a threefactor structure among the eight dependent variables. As indicated by a scree
plot, this threefactor solution accounted for approximately 81% of the shared variance. The rotated structure matrix (Table 2) revealed three factors: (a) Factor 1
consisted of the four dependent variables from the GoStop task, (b) Factor 2 consisted of the two dependent variables from the TCIP task, and (c) Factor 3 consisted of
the two dependent variables from the IMT/DMT task. The factor correlation matrix was r = .00 (Factors 1 & 2), r = .32 (Factors 1 & 3), and r = .11 (Factors 2 & 3).
Study 2
This study was an analysis of task comparison data gathered specifically to replicate and extend the results of Study 1 in a sample of adult control participants.
Again, using factor analytical methodology, our purpose was to determine whether data from the IMT/DMT, GoStop, and Two Choice impulsivity measures assess independent
aspects of laboratory-measured impulsivity.
Materials and Methods Participants—One hundred ninety-eight men and women were recruited to participate in a study of behavioral impulsivity. Potential participants
responded to advertisements for healthy adult research volunteers from 18 through 45 years old for participation in behavioral testing at Wake Forest University Health
Sciences (Winston-Salem, NC). Screening methods and inclusion/exclusion criteria were the same as those used for Study 1. Prior to participation, informed consent was
obtained from all study volunteers. This study and its procedures were reviewed and approved by the institutional review board of Wake Forest University Health
Sciences.
Procedure—Procedures and impulsivity measures, the IMT/DMT, GoStop, and TCIP, used in this second study were the same as described in Study 1. The only difference in
procedures was that the delay-reward contingencies of the TCIP were adjusting rather than fixed. Initially, the participant chose between clicking on a circle to earn
5 points after waiting 5 sec and clicking on a square to earn 15 points after waiting 15 sec. The delay contingency was adjusted dependent on the participant’s most
recent response; each short delay response decreased the length of the next long delay by 2 sec (to a minimum of 7 sec), and each long delay response increased the
length of the next long delay by 2 sec.
Data Analyses—As was the case in Study 1, performance on the behavioral impulsivity measures was analyzed using Pearson product-moment correlations and factor analysis
to evaluate the correlational structure among the eight impulsivity variables obtained from the tasks
Results
Participant Characteristics—A total of 115 men and 83 women (n = 198) completed this study. On average, participants were 28.0 years old (SD = 6.5) with an average
education of 14.2 years (SD = 2.2). Participants were of average intelligence (WASI Full Scale IQ: M = 102.0, SD = 13.6), ethnically diverse (45% African American, 51%
Caucasian, 1.5% Hispanic, and 2.5% other/multi-ethnic) and of low to middle socioeconomic status (determined by the Four Factor Index of Social Status: M = 36.2, SD =
11.2, n = 204).54
Behavioral Task Performance—Means for the IMT and DMT, ratios (commission errors/ correct detections) were .34 (SD = .17) and .35 (SD = .20), respectively. For the
GoStop, ratios averaged .07 (SD = .16), .24 (SD = .23), .49 (SD = .27) and .78 (SD = .20) for the 50, 150, 250
Dougherty et al. Page 7
Addict Disord Their Treat. Author manuscript; available in PMC 2010 June 8.
NIH-PA Author ManuscriptNIH-PA Author ManuscriptNIH-PA Author Manuscript
and 350 msec stop-delays, respectively. For the TCIP, the average of the proportion of short choices was .40 (SD = .24) and the average number of consecutive long
choices was 16.6 (SD = 14.8).
Correlations: Pearson product-moment correlations were computed among the eight dependent variables to determine the extent to which the dependent variables from the
behavioral measures were related (see Table 3). The highest correlations found were between dependent variables within each task type. For example, the IMT ratio
correlated highly with the DMT ratio. Similarly, the GoStop ratios were highly correlated with each other, as were the TCIP proportion of short choices and most
consecutive long choices. There were smaller, but statistically significant, correlations between IMT/DMT ratios and several of the GoStop ratios.
Factor Analysis: Replicating the results of Study 1, the scree plot obtained from a principal axis factor analyses revealed a three-factor structure among the eight
dependent variables. This three-factor solution accounted for approximately 79% of the shared variance. As in Study 1, the rotated structure matrix (Table 4) revealed
three factors: (a) Factor 1 consisted of the four dependent variables from the GoStop task, (b) Factor 2 consisted of the two dependent variables from the TCIP task,
and (c) Factor 3 consisted of the two dependent variables from the IMT/ DMT task. The factor correlation matrix was r = .04 (Factors 1 & 2), r = .27 (Factors 1 & 3),
and r = .09 (Factors 2 & 3).
Discussion The purpose of these two studies was to empirically test whether a number of dependent variables obtained from three laboratory behavioral measures could be
distinguished as unique components, or whether they represented only a unitary aspect of behavior. As hypothesized, results from both studies yielded three distinct
factors consisting of dependent variables from each task type, with limited inter-correlation between factors. These findings are consistent with converging evidence
from our laboratory and others that indicate behavioral impulsivity is a complex construct with multiple components. The consistency of both sets of results from our
two large samples of healthy controls, yielding three distinct factors, supports the conclusion that the three types of laboratory measures used here assess largely
distinct aspects of impulsive behavior.
The benefit of assessing distinct components of impulsive behavior is that administering these measures in combination may yield distinct patterns of performance for
particular populations or following particular treatments. For instance, a recently completed study in our laboratory55 showed a differential pattern of performance on
the three behavioral tasks (described in the present study) following alcohol administration. In this recently completed study, impulsive responding on the continuous
performance test was increased from baseline for a 2-hour period following the highest dose of alcohol (i.e., 0.8 g/kg), while impulsive responding on the stop-signal
task was increased following all doses of alcohol (i.e., 0.0, 0.2, 0.4, 0.6, and 0.8 g/kg); furthermore, impulsive responding on the delay-discounting task was
increased at both the peak and the descending limbs of the BrAC curve following the two highest alcohol doses (i.e., 0.6, and 0.8 g/kg). In contrast, acute marijuana
administration produced increased impulsive responding on a stop-signal task, but not on continuous performance or delay-discounting tasks.56 These divergent findings
support the idea of a differential pattern of performance on the tasks that may be dependent on the type of drug, as well as the need for, and importance of, using
multiple measures concurrently for identifying different component processes of behavioral impulsivity when characterizing acute drug and medication effects.
Dougherty et al. Page 8
Addict Disord Their Treat. Author manuscript; available in PMC 2010 June 8.
NIH-PA Author ManuscriptNIH-PA Author ManuscriptNIH-PA Author Manuscript
The empirical demonstration of at least three independent factors comprising behavioral impulsivity is an important addition to the literature, and extends recent
findings supporting the need for multiple measures of impulsivity.15–16, 57 Although findings from these and other recent studies suggest that behavioral impulsivity
is likely to have more than one component, a criticism of previous research has been the tendency to measure single aspects of impulsive behavior12, often in
relatively small samples. The lack of theory-driven investigation, impulsivity paradigms that measure only one component of impulsivity, and small samples, as well as
other methodological differences have prevented replication of results, thereby stirring debate regarding which methods and tasks should be used to test specific
representative model(s) of behavioral impulsivity. In contrast, the current studies compared multiple measures in two large and diverse samples of healthy adolescents
and adults and demonstrated that these tasks measured three independent factors. Having provided data in support of different types of behavioral tasks that yield
three distinct factors, our current results provide a useful foundation for future research, for validating these behavioral constructs in other populations, for
assessing the relationship of different components of impulsivity to various psychopathologies like substance use disorders, and to predict how impulsivity affects
normal development.
The use of multiple behavioral measures of impulsivity has practical and theoretical implications beyond the use of these measures for characterizing impulsivity in
particular populations or drug effects. For example, little is known about how impulsivity is related to early life-time risk for developing various substance use
disorders.58 Second, understanding of the relationship of impulsivity to early-onset substance abuse is important for understanding the affective, cognitive,
behavioral, and gene-environment mechanisms underlying adolescentonset substance use disorders.59–63 Research along these lines would inform the design of preventive
interventions to that could delay or prevent early-onset substance abuse and related impulse disorders, as well as ameliorate other psychiatric comorbidities.
To conclude, the present report supports our assertion that the most commonly used types of laboratory impulsivity tasks assess distinct aspects of behavior with
limited overlap. As a result, researchers should either use multiple behavioral measures to assess impulsivity, or use a behavioral measure likely to assess the
underlying impulsivity process of interest. While selfreport measures appear to assess independent aspects of impulsivity, these measures are useful for distinguishing
global personality traits, but they are less sensitive to behavioral fluctuations (state changes). Compared to trait measures of impulsivity, behavioral measures of
impulsivity may be useful for examining changes across time that are affected by substance use and treatment. By targeting these specific processes of interest,
researchers may more effectively identify target processes for intervention. Future studies should examine the relationship between the components of behavioral
impulsivity and the underlying developmental, behavioral, genetic, and brain mechanisms involved in the onset and maintenance of substance use disorders.
Acknowledgments
Supported by NIH Grants: R01-AA012046, R01-AA014988, R01-MH065566, R01-MH063908, R01-MH077684 and “The William & Marguerite Wurzbach Distinguished Professorship”
References
1. Moeller FG, Dougherty DM. Impulsivity and substance abuse: What is the connection? Addict Disord Their Treatment 2002;1:2–10. 2. Cloninger CR, Sigvardsson S, Bohman
M. Childhood personality predicts alcohol abuse in young adults. Alcohol Clin Exp Res 1988;12:494–505. [PubMed: 3056070] Dougherty et al. Page 9
Addict Disord Their Treat. Author manuscript; available in PMC 2010 June 8.
NIH-PA Author ManuscriptNIH-PA Author ManuscriptNIH-PA Author Manuscript
3. Elkins IJ, McGue M, Iacono WG. Prospective effects of attention/hyperactivity disorder, conduct disorder, and sex on adolescent substance use and abuse. Arch Gen
Psychiatry 2007;64:1145–1152. [PubMed: 17909126] 4. Nock MK, Kazdin AE, Hiripi E, et al. Lifetime prevalence, correlates, and persistence of oppositional defiant
disorder: results from the National Comorbidity Survey Replication. J Child Psychol Psychiatry 2007;48:703–713. [PubMed: 17593151] 5. Regier DA, Farmer ME, Rae DS, et
al. Comorbidity of mental disorders with alcohol and other drug abuse. Results from the Epidemiologic Catchment Area (ECA) Study. JAMA 1990;264:2511–2518. [PubMed:
2232018] 6. Bjork JM, Hommer DW, Grant SJ, et al. Impulsivity in abstinent alcohol-dependent patients: relation to control subjects and type 1-/type 2-like traits.
Alcohol 2004;34:133–150. [PubMed: 15902907] 7. Bickel WK, Odum AL, Madden GJ. Impulsivity and cigarette smoking: delay discounting in current, never, and ex-smokers.
Psychopharmacology (Berl) 1999;146:447–454. [PubMed: 10550495] 8. Kirby KN, Petry NM, Bickel WK. Heroin addicts have higher discount rates for delayed rewards than
non-drug-using controls. J Exp Psychol Gen 1999;128:78–87. [PubMed: 10100392] 9. Moeller FG, Dougherty DM, Barratt ES, et al. Increased impulsivity in cocaine
dependent subjects independent of antisocial personality disorder and aggression. Drug Alcohol Depend 2002;68:105– 111. [PubMed: 12167556] 10. Moeller FG, Dougherty
DM, Steinberg JL, et al. Heavy “Ecstasy” use is associated with increased impulsivity. Addict Disord Their Treatment 2002;1:47–52. 11. Quednow BB, Kuhn KU, Hoppe C, et
al. Elevated impulsivity and impaired decision-making cognition in heavy users of MDMA (“Ecstasy”). Psychopharmacology (Berl) 2007;189:517–530. [PubMed: 16425060] 12.
Moeller FG, Dougherty DM, Barratt ES, et al. The impact of impulsivity on cocaine use and retention in treatment. J Subst Abuse Treat 2001;21:193–198. [PubMed:
11777668] 13. Dawe S, Gullo MJ, Loxton NJ. Reward drive and rash impulsiveness as dimensions of impulsivity: Implications for substance misuse. Addict Behav
2004;29:1389–1405. [PubMed: 15345272] 14. Gorlyn M, Keilp JG, Tryon WW, et al. Performance test correlates of component factors of impulsiveness. Pers Individ Dif
2005;38:1549–1559. 15. Lane SD, Cherek DR, Rhoades HM, et al. Relationships among laboratory and psychometric measures of impulsivity: Implications in substance abuse
and dependence. Addict Disord Their Treatment 2003;2:33–40. 16. Mitchell SH. Measures of impulsivity in cigarette smokers and non-smokers. Psychopharmacology (Berl)
1999;146:455–464. [PubMed: 10550496] 17. Reynolds B, Ortengren A, Richards JB, et al. Dimensions of impulsive behavior: Personality and behavioral measures. Pers
Individ Dif 2006;40:305–315. 18. Whiteside SP, Lynam DR. The five factor model and impulsivity: using a structural model of personality to understand impulsivity. Pers
Individ Dif 2001;30:669–689. 19. Zelenski JM, Larsen RJ. Susceptibility to affect: A comparison of three personality taxonomies. J Pers 1999;67:761–791. [PubMed:
10540757] 20. Miller E, Joseph S, Tudway J. Assessing the component structure of four self-report measures of impulsivity. Pers Individ Dif 2004;37:349–358. 21. Dawe
S, Loxton NJ. The role of impulsivity in the development of substance use and eating disorders. Neurosci Biobehav Rev 2004;28:343–351. [PubMed: 15225976] 22. Franken
IH, Muris P, Georgieva I. Gray’s model of personality and addiction. Addict Behav 2006;31:399–403. [PubMed: 15964149] 23. Moeller FG, Barratt ES, Dougherty DM, et al.
Psychiatric aspects of impulsivity. Am J Psychiatry 2001;158:1783–1793. [PubMed: 11691682] 24. Dougherty DM, Marsh DM, Mathias CW, et al. The conceptualization and
role of impulsivity: Bipolar Disorder and substance abuse. Psychiatric Times 2005;22:32–35. 25. Rosvold HE, Mirsky AF, Sarason I, et al. A continuous performance test
of brain damage. J Consult Psychol 1956;20:343–350. [PubMed: 13367264] Dougherty et al. Page 10
Addict Disord Their Treat. Author manuscript; available in PMC 2010 June 8.
NIH-PA Author ManuscriptNIH-PA Author ManuscriptNIH-PA Author Manuscript
26. Halperin JM, Wolf LE, Pascualvaca DM, et al. Differential assessment of attention and impulsivity in children. J Am Acad Child Adolesc Psychiatry 1988;27:326–329.
[PubMed: 3379014] 27. Halperin JM, Wolf L, Greenblatt ER, et al. Subtype analysis of commission errors on the Continuous Performance Test. Dev Neuropsychol
1991;7:207–217. 28. Sostek AJ, Buchsbaum MS, Rapoport JL. Effects of amphetamine on vigilance performance in normal and hyperactive children. J Abnorm Child Psychol
1980;8:491–500. [PubMed: 7462528] 29. Sykes DH, Douglas VI, Morgenstern G. Sustained attention in hyperactive children. J Child Psychol Psychiatry 1973;14:213–220.
[PubMed: 4750855] 30. Wohlberg GW, Kornetsky C. Sustained attention in remitted schizophrenics. Arch Gen Psychiatry 1973;28:533–537. [PubMed: 4692152] 31. Logan, GD.
On the ability to inhibit thought and action: A users’ guide to the Stop Signal Paradigm. In: Dagenbach, D.; Carr, TH., editors. Inhibitory processes in attention,
memory, and language. Academic Press; San Diego, CA: 1994. p. 189-239. 32. Dougherty, DM.; Marsh, DM. IMT/DMT & Delayed Memory Tasks: A research tool for studying
attention, memory, and impulsive behavior [Computer software and manual]. Version 2. Neurobehavioral Research Laboratory and Clinic; San Antonio, TX: 2003. 33.
Dougherty DM, Bjork JM, Harper RA, et al. Validation of the Immediate and Delayed Memory Tasks in hospitalized adolescents with disruptive behavior disorders. Psychol
Rec 2003;53:509–532. 34. Dougherty, DM.; Mathias, CW.; Marsh, DM. Impulsivity and aggression in adolescents with Conduct Disorder. In: Barratt, ES., editor. Clinical
Research on Impulsivity: New Developments and Directions for Possible Treatments; Symposium conducted at the World Psychiatric Association Congress (Zone 8); Florence,
Italy. 2004. chair 35. Dougherty DM, Mathias CW, Marsh DM, et al. Laboratory measured behavioral impulsivity relates to suicide attempt history. Suicide Life Threat
Behav 2004;34:374–385. [PubMed: 15585459] 36. Dougherty DM, Marsh DM, Mathias CW, et al. The effects of alcohol on laboratory-measured impulsivity after L-tryptophan
depletion or loading. Psychopharmacology (Berl) 2007;193:137–150. [PubMed: 17377773] 37. Mathias CW, Dougherty DM, Marsh DM, et al. Laboratory measures of impulsivity:
A comparison of women with and without childhood aggression. Psychol Rec 2002;52:289–303. 38. First, MB.; Spitzer, RL.; Gibbon, M., et al. Structured Clinical
Interview for DSM-IV-TR Axis I Disorders, Research Version, Non-patient Edition (SCID-I/NP). Biometrics Research, New York State Psychiatric Institute; New York: 2001.
39. Kaufman J, Birmaher B, Brent D, et al. Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present and Lifetime Version (K-SADS-PL): Initial
reliability and validity data. J Am Acad Child Adolesc Psychiatry 1997;36:980–988. [PubMed: 9204677] 40. Shaffer D, Fisher P, Lucas C, et al. NIMH Diagnostic Interview
Schedule for Children Version IV (NIMH DISC-IV): Description, differences from previous versions and reliability of some common diagnoses. J Am Acad Child Adolesc
Psychiatry 2000;39:28–38. [PubMed: 10638065] 41. Psychological Corporation. Wechsler Abbreviated Scale of Intelligence® (WASI) Manual. Harcourt Brace and Company; San
Antonio, TX: 1999. 42. Dougherty DM, Marsh DM, Mathias CW. Immediate and Delayed Memory Tasks: A computerized measure of memory, attention, and impulsivity. Behav Res
Methods Instrum Comput 2002;34:391– 398. [PubMed: 12395555] 43. Dougherty DM, Marsh DM, Moeller FG, et al. Effects of moderate and high doses of alcohol on attention,
impulsivity, discriminability, and response bias in Immediate and Delayed Memory Task performance. Alcohol Clin Exp Res 2000;24:1702–1711. [PubMed: 11104118] 44.
Dougherty DM, Bjork JM, Harper RA, et al. Behavioral impulsivity paradigms: A comparison in hospitalized adolescents with disruptive behavior disorders. J Child
Psychol Psychiatry 2003;44:1145–1157. [PubMed: 14626456] 45. Dougherty DM, Mathias CW, Marsh DM, et al. Laboratory measured behavioral impulsivity relates to suicide
attempt history. Suicide Life Threat Behav 2004;34:374–385. [PubMed: 15585459] 46. Swann AC, Anderson J, Dougherty DM, et al. Measurement of interepisode impulsivity
in bipolar disorder. Psychiatry Res 2001;101:195–197. [PubMed: 11286822] Dougherty et al. Page 11
Addict Disord Their Treat. Author manuscript; available in PMC 2010 June 8.
NIH-PA Author ManuscriptNIH-PA Author ManuscriptNIH-PA Author Manuscript
47. Swann AC, Bjork JM, Moeller FG, et al. Two models of impulsivity: Relationship to personality traits and psychopathology. Biol Psychiatry 2002;94:988–994. [PubMed:
12062883] 48. Dykman RA, Ackerman PT, Oglesby DM. Selective and sustained attention in hyperactive, learningdisabled and normal boys. J Nerv Ment Dis 1979;167:288–297.
[PubMed: 448331] 49. Dougherty, DM.; Mathias, CW.; Marsh, DM. GoStop Impulsivity Paradigm [Manual].Version 1. Neurobehavioral Research Laboratory and Clinic; San
Antonio, TX: 2003. 50. Dougherty DM, Mathias CW, Marsh DM, et al. Laboratory behavioral measures of impulsivity. Behav Res Methods 2005;37:82–90. [PubMed: 16097347] 51. Marsh DM, Dougherty DM, Mathias CW, et al. Comparison of women with high and low trait impulsivity using laboratory impulsivity models of response-disinhibition
and reward-choice. Pers Individ Dif 2002;33:1291–1310. 52. Dougherty, DM.; Marsh, DM.; Mathias, CW. Two Choice Impulsivity Paradigm [Computer software and manual].
Version 1. Neurobehavioral Research Laboratory and Clinic; San Antonio, TX: 2003. 53. Rachlin, H. The science of self control. Harvard University Press; Boston, MA:
2000. 54. Hollingshead, AB. Four Factor Index of Social Status. Department of Sociology, Yale University; New Haven, CT: 1975. 55. Dougherty, DM.; Marsh-Richard, DM.;
Hatzis, ES., et al. Alcohol’s dose-dependent effects on multiple behavioral measures of impulsivity across the breath alcohol concentration curve. In Review 56.
McDonald J, Schleifer L, Richards JB, et al. Effects of THC on behavioral measures of impulsivity in humans. Neuropsychopharmacology 2003;28:1356–1365. [PubMed:
12784123] 57. Monterosso J, Ehrman R, Napier KL, et al. Three decision-making tasks in cocaine-dependent patients: do they measure the same construct? Addiction
2001;96:1825–1837. [PubMed: 11784475] 58. Kessler RC, Amminger GP, Aguilar-Gaxiola S, et al. Age of onset of mental disorders: A review of the literature. Curr Opin
Psychiatry 2007;20:359–364. [PubMed: 17551351] 59. Caspi A, Sugden K, Moffitt TE, et al. Influence of life stress on depression: Moderation by a polymorphism in the 5
-HTT gene. Science 2003;18:386–389. [PubMed: 12869766] 60. Chambers RA, Taylor JR, Potenza MN. Developmental neurocircuitry of motivation in adolescence: A critical
period of addiction vulnerability. Am J Psychiatry 2003;160:1041–1052. [PubMed: 12777258] 61. Chassin L, Presson CC, Rose J, et al. What is addiction? Age-related
differences in the meaning of addiction. Drug Alcohol Depend 2007;87:30–38. [PubMed: 16930860] 62. Clark DB, Jones BL, Wood DS, et al. Substance use disorder
trajectory classes: Diachronic integration of onset age, severity, and course. Addict Behav 2006;31:995–1009. [PubMed: 16675151] 63. Dawes MA, Antelman SM, Vanyukov
MM, et al. Developmental sources of variation in liability to adolescent substance use disorders. Drug Alcohol Depend 2000;61:3–14. [PubMed: 11064179] Dougherty et al. Page 12
Addict Disord Their Treat. Author manuscript; available in PMC 2010 June 8.
NIH-PA Author ManuscriptNIH-PA Author ManuscriptNIH-PA Author Manuscript
Figure 1.
Dougherty et al. Page 13
Addict Disord Their Treat. Author manuscript; available in PMC 2010 June 8.
NIH-PA Author ManuscriptNIH-PA Author ManuscriptNIH-PA Author Manuscript
NIH-PA Author ManuscriptNIH-PA Author ManuscriptNIH-PA Author Manuscript
Dougherty et al. Page 14
Table 1 Study 1: Correlations between Laboratory Measures of Impulsivity (n = 204)
DMT RatioGoStop Ratio 50 msec
GoStop Ratio 150 msec
GoStop Ratio 250 msec
GoStop Ratio 350 msec
TCIP Proportion ShortTCIP Most Consecutive Long
Dependent Variablesrrrrrrr IMT Ratio.71**.11.21*.31**.32**−.12.09 DMT Ratio—.09.15.26**.25**−.13−.07 GoStop Ratio 50 msec—.74**.53**.36**.04−.07 GoStop Ratio 150
msec—.78**.59**−.10.01 GoStop Ratio 250 msec—.78**−.02−.05 GoStop Ratio 350 msec—−.03−.07 TCIP Proportion Short Responses—−.81**
*
indicates significance < .05, ** indicates significance < .001
Addict Disord Their Treat. Author manuscript; available in PMC 2010 June 8.
NIH-PA Author ManuscriptNIH-PA Author ManuscriptNIH-PA Author Manuscript
Dougherty et al. Page 15
Table 2 Study 1: Factor Structure of Three Laboratory Behavioral Impulsivity Measures (n = 204)
Factor
Dependent Variables 1 2 3 IMT Ratio .287 .111 .867* DMT Ratio .224 .105 .809* GoStop Ratio 50 msec .659* −.045 .102 GoStop Ratio 150 msec .937* .069 .204 GoStop Ratio
250 msec .897* −.019 .374 GoStop Ratio 350 msec .706* −.027 .370 TCIP Proportion of Short Choices −.040 .904* −.142 TCIP Consecutive Long Choices −.048 −.892*† .081
*
denotes factor loading > |.6|
†
denotes negative factor loading reflecting fewer TCIP Consecutive Long Choices, indicating greater impulsivity.
Addict Disord Their Treat. Author manuscript; available in PMC 2010 June 8.
NIH-PA Author ManuscriptNIH-PA Author ManuscriptNIH-PA Author Manuscript
Dougherty et al. Page 16
Table 3 Study 2: Correlations between Laboratory Measures of Impulsivity (n = 198)
DMT RatioGoStop Ratio 50 msec
GoStop Ratio 150 msec
GoStop Ratio 250 msec
GoStop Ratio 350 msec
TCIP Proportion ShortTCIP Most Consecutive Long
Dependent Variablesrrrrrrr IMT Ratio.68**.02.23**.25**.16*.00−.06 DMT Ratio—.05.24**.20*.10.13−.14 GoStop Ratio 50 msec—.72**.50**.32**.04−.04 GoStop Ratio 150
msec—.74**.50**.05−.03 GoStop Ratio 250 msec—.65**.04.01 GoStop Ratio 350 msec—.02.00 TCIP Proportion Short Responses—−.82**
*
indicates significance < .05, ** indicates significance < .001
Addict Disord Their Treat. Author manuscript; available in PMC 2010 June 8.
NIH-PA Author ManuscriptNIH-PA Author ManuscriptNIH-PA Author Manuscript
Dougherty et al. Page 17
Table 4 Study 2: Factor Structure of Three Laboratory Behavioral Impulsivity Measures (n = 198)
Factor
Dependent Variables 1 2 3 IMT Ratio .222 .026 .903* DMT Ratio .199 .143 .764* GoStop Ratio 50 msec .669* .045 .046 GoStop Ratio 150 msec .927* .047 .278 GoStop Ratio
250 msec .846* .012 .289 GoStop Ratio 350 msec .604* .004 .181 TCIP Proportion of Short Choices .049 .919* .070 TCIP Consecutive Long Choices −.021 −.895*† −.108
*
denotes factor loading > |.6|
†
denotes negative factor loading reflecting fewer TCIP Consecutive Long Choices, indicating greater impulsivity.
Addict Disord Their Treat. Author manuscript; available in PMC 2010 June 8.

TO GET YOUR ASSIGNMENTS DONE AT A CHEAPER PRICE,PLACE YOUR ORDER WITH US NOW