Experimental Research on Jury Decision-Making

Author(s): Robert J. MacCoun

Source: Science, New Series, Vol. 244, No. 4908 (Jun. 2, 1989), pp. 1046-1050
Published by: American Association for the Advancement of Science
Stable URL: https://www.jstor.org/stable/1703992
Accessed: 07-02-2019 03:49 UTC

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/1703992?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

American Association for the Advancement of Science is collaborating with JSTOR to

digitize, preserve and extend access to Science

This content downloaded from 49.150.13.208 on Thu, 07 Feb 2019 03:49:05 UTC
All use subject to https://about.jstor.org/terms
 Experimental Research on Jury Decision-Making
ROBERT J. MACCOUN

researchers would like controlled experiments with random assign-

Because trial juries deliberate in secrecy, legal debates ment to conditions under study. In some instances, courts have
about jury functioning have relied heavily on anecdote randomly assigned jury trials to alternative procedures, but manipu-
and speculation. In recent years, investigators have begun lations of many variables of interest are not ethically or legally
to challenge many common assumptions about jury be- feasible in actual trial settings (6). Mock juries must be used for most
havior. An important tool in this effort has been the mock experiments; in these, research participants are asked to reach
jury experiment, in which research participants are ran- judgments regarding a simulated legal trial. The mock jury approach
domly assigned to alternative trial conditions and asked has the added advantage of permitting replication across juries
to reach a verdict in a simulated case. Researchers have within the context of a single trial, and there is no legal barrier to
used mock jury experiments to test hypotheses about observing deliberation (7).
causal influences on jury behavior and to develop theoret- Differences between these experiments and actual trials have led
ical models of the jury deliberation process. some observers to question whether generalizable conclusions about
actual jury behavior can be reached by studying the behavior of
mock juries reacting to written, audiotaped, or videotaped trial
reenactments (8). The effects of several factors that distinguish mock
JURY VERDICTS DIRECTLY AFFECT THE LIVES OF HUNDREDS OF jury simulations from actual trials have been assessed empirically.
thousands of individuals in the United States every year and Experiments comparing mock jurors with subjects who thought
serve a broader bellwether function in plea bargaining and they were actually trying a case have been inconclusive; different
settlement negotiations (1). But because jury deliberation is cloaked studies have found mock jurors' verdicts to be more lenient, less
in secrecy, legal policy-makers have made important decisions about lenient, and no different from those of "actual" jurors (9). Other
the scope and conduct of jury trials on the basis of untested studies have examined the effects of the frequent use of college
intuitions about how juries reach their verdicts (2, 3). In this review
students as mock jurors, finding little or no difference in compari-
of research on jury behavior, I will emphasize the use of mock jury sons of verdicts by student and adult jury-eligible respondents for
experiments to test hypotheses and refine theoretical models of the the same cases (10, 11). There is some evidence that simulated trial
decision process. Because jury decision-making involves two differ- presentations might artificially exaggerate the impact of experimen-
ent phases-cognitive processing during the trial and deliberation in tally manipulated variables, particularly defendant characteristics
the jury room-I review research on both the trial and deliberation (12). But mock jurors do not appear to reach decisions by a
phases of the judgment process. In keeping with the emphasis of fundamentally different process than actual jurors (8, 13). When the
most jury research, I focus primarily on decision-making in criminal objective is to precisely estimate the magnitude of relations among
trials; the extent to which these findings generalize to civil litigation
variables in actual jury trials, the archival method is more appropri-
is not clear (3). ate. The role of mock jury experimentation is to explain the
processes underlying those relations.

Jury Research Methods

Predeliberation Juror Judgment
In the 1950s, researchers at the University of Chicago covertly
recorded the deliberations of several federal juries. Despite the Evidence evaluation at trial. One of the earliest findings in mock jury
court's cooperation, this endeavor was aborted by a congressional research was that despite judges' instructions to the contrary, many
inquiry, resulting in legislation prohibiting attempts to observe or jurors form tentative verdict preferences early in the trial (14), a
record jury deliberation (4). Since then, researchers have resorted to finding that underscores the importance of studying predeliberation
other strategies to study jury behavior, most notably archival juror judgment as an adjunct to research on jury deliberation.
analyses and mock jury experiments. In the archival approach, jury Rather than cataloging the dozens of trial and case characteristics
verdicts are sampled from court records or court reporting services that have been found to influence jurors, I will briefly review several
and analyzed statistically to describe longitudinal trends and to theoretical models of individual juror judgment that have proved
identify relations between verdicts and case characteristics (5). But useful for stimulating research on general principles of juror judg-
archival data sources omit a great deal of potentially relevant ment (15).
information and only document what juries have done, not how or The Bayesian and information integration models each represent a
why they did it. Researchers must attempt to infer the latter, which juror's evaluation of the evidence as a unidimensional subjective
is precarious because one can never completely disentangle the probability judgment. These models are framed in mathematical
natural covariation among various case and trial characteristics.
The author is a social psychologist in the Behavioral Sciences Department at The RAND
In order to better understand the jury decision-making process, Corporation, Santa Monica, CA 90406-2138.

io46 SCIENCE, VOL. 244

This content downloaded from 49.150.13.208 on Thu, 07 Feb 2019 03:49:05 UTC
All use subject to https://about.jstor.org/terms
terms, but theorists are not proposing that jurors literally carry out knowledge structures abstracted from prior experience-to rem
such calculations in their heads: "now I'll divide by 0.5 and carry the ber and organize trial evidence into a plausible story. Jurors t
I." Instead, the formulas are a useful way of explicitly predicting the attempt to match the story to available verdict categories, sel
functional relations among a juror's assessment of each evidentiary the verdict that provides the best fit. The story model is a psych
item and his or her overall judgment. In the simplest Bayesian model
ically plausible account of juror decisionmaking, and it is th
(15), the juror's estimate of the odds of guilt given n items of trial
model in which serious consideration is given to the role of me
evidence is Rn = Ro(L1)(L2), . .., (L,) where Ro = P(Gleo)/ processes during the trial, but more research is needed to estab
P(NGIeo), which represents the juror's initial odds estimate of the
its predictive validity and heuristic value for generating testa
defendant's being guilty (G) or not guilty (NG) given the fact that
hypotheses.
the defendant is on trial (eo), and Li - P(eilG)/P(eilNG), a likelihood
juror biases. Intuition and courtroom folklore suggest that j
ratio representing the perceived diagnosticity of item e. More personal characteristics might predispose them toward certain ver-
sophisticated versions use complex hierarchical chains of inference dicts. Attorneys attempt to detect these predispositions during jury
to account for dependencies among evidentiary items (16). Argu- selection proceedings, traditionally relying on hunches and stereo-
ably, Bayesian models depict how a "rational" person would aggre- typic rules of thumb. In recent years, some defense lawyers have
gate the evidence, but they generally do a poor job of describing hired social scientists to conduct "scientific jury selection," which
juror judgment (15). For example, relative to Bayesian norms, mockusually involves a survey of community knowledge and attitudes
jurors do not adequately adjust their judgments to take into account regarding the issues in dispute, occasionally supplemented by
forensic incidence statistics presented in expert testimony-for clinical observation of potential jurors under questioning (25). The
instance, the likelihood that the offender and the defendant wouldrelation of survey items intended to serve as a proxy for verdict
have matching hair samples by chance (17). Also, mock jurors havepreference with various demographic, personality, and attitudinal
been found to "double count" redundant testimony from corrobora- variables is evaluated by regression analysis in order to build a
tive witnesses (16). statistical profile of the clients ideal juror.
The information integration model of juror judgment is ground- However, a large body of empirical research calls into question
ed in basic research on human judgment (18). According to this the premise that jurors' votes during deliberation can be reliably
model, a juror's evaluation of the evidence, J, can be described by apredicted from juror characteristics that are observable before trial.
weighted average of the pretrial opinion, so, and the subjective In general, jurors' demographic attributes, personality traits, and
probability of guilt or liability implied by each evidentiary item, Si. general attitudes are associated weakly and unreliably with jurors'
Thus, J = (woso + Ywisi)l(wo + Ywi), where wo and wi refer to the verdicts (1, 13, 26). For example, in a study of over 800 mock jurors
weight given to the pretrial opinion and the ith evidentiary item, recruited from Boston-area jury pools, jurors' education, occupa-
respectively (15, 19). These weights are postulated to reflect each tion, political ideology, gender, age, and trial experience collectively
item's perceived relevance and credibility. Psychologically, this accounted for less than 2% of the variance in their verdict prefer-
model can be characterized as a process in which the juror's global ences (13). But proponents of scientific jury selection argue that the
judgment is continually adjusted so as to fall between its previous approach is most effective in trials involving controversial issues.
value and the value of each new piece of evidence that is presented at Capital punishment might be one such issue; numerous studies have
trial (20). Mock juror research findings are generally consistent withfound that attitudes toward the death penalty in the abstract reliably
this averaging model (19), although rigorous tests are rare (15). This predict the decision to vote to convict a defendant accused of
model can account for the underuse of incidence statistics and other homicide (27).
forms of base-rate evidence; they are simply "averaged in" along A more robust source of bias in juror judgment results from
with other less statistically reliable items of evidence (20). The modelexposure to extralegal information. Studies have documented reli-
can also explain the impact of redundant testimony; by attenuating able effects of pretrial publicity, inadmissible evidence, and litigants'
the contribution of the pretrial opinion, two items with the same physical characteristics on mock juror judgments (1, 28, 29). When
value can produce a more extreme judgment than either one alone seemingly probative information is ruled inadmissible because of
(18). due process violations, jurors may nevertheless incorporate it in their
The Bayesian and information integration models represent the probability-of-guilt assessment (29). But other extralegal factors,
juror's judgment on a continuous scale, but jurors are typically such as the attractiveness of an automobile theft victim, appear to
required to reach a categorical verdict. Thus, the juror's judgment influence verdicts indirectly by heightening the anticipated regret of
must be compared to a decision threshold, which should ideally either convicting the innocent or acquitting the guilty and thereby
correspond to the assigned standard of proof; for example, the raising or lowering jurors' standard-of-proof thresholds (30).
criminal "reasonable doubt" standard. But because legal definitions
of these standards are ambiguous, there is considerable variance in
the thresholds jurors actually adopt (21, 22). Several theorists (23)
have suggested a decision-theory derivation of the juror's threshold.
The Deliberation Process
Given the perceived probability of guilt, p, and the expected regret Charting factional movement in the jury. In the 1960s, Kalven and
over convicting an innocent defendant, D~i, or acquitting a guilty Zeisel (31) used post-trial juror interviews to reconstruct the initial
defendant, Dag, the juror can minimize his or her expected regret by ballots in 225 criminal jury deliberations. Of the 146 juries with a
setting a threshold value of p* D~i/(D~i + Dag). With this model,nonunanimous majority at first ballot, only seven reached the verdict
it is possible to predict jurors' verdicts about 80% of the time by advocated by the minority faction. Kalven and Zeisel suggested that
matching their probability-of-guilt ratings to the value of p* esti-"with very few exceptions the first ballot decides the outcome of the
mated from their expected regret ratings (21, 22). This is a better verdict. . .. The deliberation process might well be likened to what
than chance rate, although it suggests that better models are needed.
the developer does for an exposed film; it brings out the picture, but
The recent "story" model (13, 15, 24) of juror cognition departs the outcome is predetermined" (31, pp. 488-489).
from these unidimensional approaches. The model is an extension of In the years since Kalven and Zeisel's analysis, a number of
basic research on the cognitive representation of narrative informa-stochastic models of jury decision-making have been devreloped (32),
tion. According to this model, jurors use episode schemata-generic some of which have been implemented as computer simulations ( 13,

2 JUNE 1989 ARTICLES J047

This content downloaded from 49.150.13.208 on Thu, 07 Feb 2019 03:49:05 UTC
All use subject to https://about.jstor.org/terms
33). A common feature of these models is the use of the group mentum-like effect in which a jury's next transition can be pred
state-the distribution of jurors across distinct verdict factions-as by the direction of its previous transition (36). Nevertheless, the
the unit of analysis. For example, an 8:4 state would indicate that 8 violations of these assumptions are of relatively small magnitude,
jurors favor conviction and 4 jurors favor acquittal in a 12-person and Markov process models have been fairly successful at predicting
jury. These models make a number of assumptions about the mock jury behavior (13, 33).
dynamics of jury deliberation that have been assessed empirically. Sources of influence. By comparing the deliberation process to the
Following Kalven and Zeisel (31), each model has as an assump- development of a photograph, Kalven and Zeisel (31) implied that
tion that a faction's influence is a function of its relative size. This jury deliberation might be a mere vote-counting formality. This is an
majority effect is easily the most robust finding in mock jury research
exaggeration. The strength-in-numbers effect of the majority in-
(13, 32). Even when a jury is ostensibly operating under a unanimity volves two different sources of influence: normative influence, the
decision rule, its verdict can be predicted fairly reliably by an implicit conformity pressures brought to bear on a minority faction, and
two-thirds majority rule. A related phenomenon, group polariza- informational influence, the number and persuasiveness of argu-
tion, has been documented in mock jury research and in hundreds of ments generated to support a position (34). In content analyses of
other small group laboratory studies (19, 34). To the extent that a deliberation, both types of influence are found and both are
group's average predeliberation opinion deviates from the neutral
correlated with mock jurors' final votes (37). Although a faction's
point on a bipolar scale, the average postdeliberation opinion will size and its ability to generate arguments are naturally confounded,
tend to be more extreme in the same direction. All other things experiments in which one source of influence is held constant while
being equal, polarization does not occur when equal sized and the other is manipulated indicate that both sources affect voting
equally opinionated factions are opposed. patterns (34).
However, in criminal juries, all other things are not equal. One Informational influence during deliberation is desirable to the
might expect influence to be symmetrical within equal-split group extent that it fosters more complete and accurate recall of trial
states (for example, 6:6) or between equal-ratio group states (for evidence and corrects errors and biases. Ideally, Pj, the probability
example, 8:4 with 4:8). If so, each faction in a 6:6 group would be that the jury will recall an item of evidence, should equal 1 -
equally likely to win, and a two-thirds majority would be as likely to
(1 - Pi) n, where Pj is the probability that an individual will recall
win no matter which verdict it favored. Nevertheless, there is it and n is the group's size, but actual group recall tends to fall
considerable evidence that factions favoring acquittal have more somewhere between this model and a "majority rule" model-that
influence than factions of equivalent size favoring conviction (11, is, social support is often needed to convince others that a recollec-
32). This asymmetry effect appears to be a consequence of the tion is correct (38). Content analyses of deliberation indicate that in
asymmetric reasonable doubt standard used in criminal trials. In a some instances jurors admonish each other to ignore inadmissible
recent experiment (11), the asymmetry effect was reproduced when evidence (29), but information integration theorists (19) have
mock juries were assigned the reasonable doubt standard, but argued that even without such admonishments, the recollection and
influence was symmetrical in mock juries that tried the same criminal discussion of trial evidence should reduce the relative weight given
case under the symmetrical "preponderance of evidence" standard to extralegal biases. Although studies in which imbalanced trial
used in civil litigation. The reasonable doubt standard appears to evidence is used show that deliberation attenuates biases (19),
provide a rhetorical advantage for advocates of acquittal during studies with very close cases show sustained or even enhanced bias
after deliberation (22, 39). Kalven and Zeisel (31) suggested that
deliberation, and the effect of the standard is thereby amplified by
group discussion. close cases might "liberate" jurors from the evidence and allow their
A common modeling assumption is that influence is proportional personal sentiments to influence their judgment.
to the relative size of a faction, but not its absolute size: for example,
that 8:4, 4:2, and 2:1 are functionally equivalent group states. In
Williams v. Florida (2), the Supreme Court explicitly adopted this
Effects of Structural Task Variables
proportionality assumption in its decision to uphold the use of 6-
person criminal juries in state courts. In an experimental comparison During the 1970s, a number of controversial Supreme Court
of 12-, 6-, and 3-person mock juries, two different violations of decisions relaxed the traditional requirement that a jury consist of 12
proportionality were observed, both of which agree with replicate members operating under a unanimous decision rule (2). In doing
basic results in conformity research (35). First, there was more so, the Supreme Court explicitly assumed that within certain limits,
majority influence in the 2:1 state than in the 4:2 or 8:4 states; that a jury's size and decision rule would not influence its functioning.
is, lone minorities fared more poorly than their proportional Since then, a considerable body of research on the effects of these
counterparts. Second, a minority-of-one was less likely to yieldvariables
to a has accumulated, much of it too late to dissuade the
2-person majority than to a 5- or 1 1-person majority. Nevertheless,
Supreme Court. Neither of these variables systematically affects
the proportionality assumption held up well in comparisons of 6- verdicts in carefully controlled mock jury experiments, but smaller
and 12-person juries, which encompasses the existing range of and nonunanimous juries recall less evidence, deliberate more
permissible jury sizes in the state and federal courts. quickly and less thoroughly, and are more likely to reach a verdict
The transition among group states during jury deliberation canthan
be their traditional counterparts (13, 35). Moreover, rudimentary
modeled as a discrete-state Markov process (33, 36) under the sampling theory indicates that a smaller jury will be less repre-
assumption that the process is both stationary and path-indepen- sentative of minority viewpoints in the community, and mathemati-
dent. Minor violations of each assumption have been found in mock cal simulations suggest that smaller and nonunanimous juries might
jury research. The process would be stationary if the probability be
of more
a likely to falsely convict the innocent or acquit the guilty
given transition between group states remained constant through- (32). Whether these structural changes have increased trial efficiency
out the deliberation process, but stronger majority and asymmetry enough to offset potentially deleterious effects on the performance
effects have been found during the second half of deliberation (36). and perceived legitimacy of the jury remains an issue for public
The process would be path-independent if the likelihood of a debate (40).
transition between states was independent of the group's history of Jury researchers have devoted the most attention to cases in which
previous transitions, but there is correlational evidence for a mo- a jury must render a dichotomous criminal verdict, guilty or not

I048 SCIENCE, VO L.244

This content downloaded from 49.150.13.208 on Thu, 07 Feb 2019 03:49:05 UTC
All use subject to https://about.jstor.org/terms
guilty. Several studies suggest that the availability of multiple KagehiroJ. Appl. Soc. Psychol. 15, 409 (1985).
6. L. Heuer and S. Penrod, Law Human Behav. 12, 231 (1988); H. Zeisel, Stanford
response options can fimdamentally alter the jury's decision. For
Law Rev. 30, 491 (1978); Experimentation in the Law (Federal Judicial Center,
example, providing mock jurors with a guilty-but-mentally-ill op- Washington, DC, 1981).

tion results in a significant reduction in not guilty by reason of 7. Posttrial juror interviews are also useful for examining how jurors reach their
decisions [C. L. Hinchcliff, Marquette Law Rev. 69, 495 (1986); M. Selvin and L.
insanity verdicts in insanity defense cases (41), and there are similar Picus, The Debate over Jury Performance: Observations from a Recent Asbestos Case
response option effects in homicide cases (42). Juries are occasionally (RAND, Santa Monica, CA, 1987)]. However, jurors' retrospective accounts are
vulnerable to memory loss and distortion, and juror interviewing also raises ethical
asked to reach multiple verdicts, as when criminal defendants are
and legal concerns [Note, Harvard Law Rev. 96, 886 (1983)]. Another approach is
tried for multiple offenses involving separate incidents. A number of to recruit "shadow jurors," citizens who sit in the courtroom and react to an
experiments have shown that mock jurors are more likely to convict ongoing trial [D. E. Vinson, Am. Bar Assoc. J. 68, 1243 (1982)]. However, the
duration of actual trials can make replication across observers or trials prohibitively
a defendant of a given charge in a joined trial than when the same
expensive. Also, since these methods are strictly correlational, neither permits
charge is tried separately (43). unambiguous causal inferences.
Because jury research has focused almost exclusively on criminal 8. Reviewed by R. M. Bray and N. L. Kerr, in The Psychology of the Courtroom, N. L.
Kerr and R. M. Bray, Eds. (Academic Press, New York, 1982), p. 287; see also E.
cases, relatively little is known about how juries allocate civil liability A. Lind and L. Walker, Law Hum. Behav. 3, 5 (1979); D. G. Mook, Am. Psychol.
among parties and assess compensatory and punitive damages, 38, 379 (1983).

continuous judgments that might evoke very different decision 9. M. F. Kaplan and S. Krupa, Law Psychol. Rev. 10, 1 (1986); N. L. Kerr, D. R.
Nerenz, D. Herrick, Sociol. Methods Res. 7, 337 (1979); D. W. Wilson and E.
processes. This is unfortunate because there is currently an active DonnersteinJ. Appl. Soc. Psychol. 7, 175 (1977); H. Zeisel and S. S. Diamond,
legal debate about the jury's role in resolving product liability, Stanford Law Rev. 30, 491 (1978).
10. R. M. Bray et al., Soc. Psychol. 41, 256 (1978); S. Sue, R. E. Smith, R. GilbertJ.
medical malpractice, and other civil disputes. There is also a growing
Crim. Justice 2, 163 (1974).
use of special interrogatories and itemized verdicts, which require 11. R. J. MacCoun and N. L. Kerr, J. Pers. Soc. Psychol. 54, 21 (1988).
jurors to disaggregate complex decisions. These topics are ripe for 12. R. Juhnke et al., J. Pers. Soc. Psychol. 5, 36 (1979); S. Tanford and S. PenrodJ.
Pers. Soc. Psychol. 47, 749 (1984).
theoretical development and research (3). 13. R. Hastie, S. D. Penrod, N. Pennington, Inside the Jury (Harvard Univ. Press,
Cambridge, 1983).
14. H. P. Weld and M. Roff, Am.J. Psychol. 51, 609 (1938); H. P. Weld and E. R.
Danzig, ibid. 53, 518 (1940); P. G. Devine and T. M. Ostrom, J. Pers. Soc. Psychol.
49, 5 (1985); S. M. Kassin and L. S. Wrightsman, ibid. 37, 1877 (1979).
Conclusions
15. These models are examined in detail by N. Pennington and R. Hastie, Psychol. Bull.
89, 246 (1981).
Empirical research on jury functioning is gradually replacing the
16. D. A. Schum, A. W. Martin, Law Soc. Rev. 17, 105 (1982).
reliance on anecdotes and speculation in the legal policy domain. 17. D. L. Faigman and A. J. Baglioni, Law Hum. Behav. 12, 1 (1988); W. C.
Much is now known about cognitive processing at trial and the Thompson and E. L. Schumann, ibid. 11, 167 (1987).
18. N. H. Anderson, Foundations of Information Integration Theory (Academic Press, New
dynamics of jury deliberation, and the effects of many key trial York, 1981).
variables are generally understood. But in an evaluation of the jury's 19. M. F. Kaplan and L. E. Miller, J. Pers. Soc. Psychol. 36, 1443 (1978).
20. L. L. Lopes, Bull. Psychonomic Soc. 23, 509 (1985).
merit as a legal institution many dimensions must be considered-
21. F. C. Dane, Law Hum. Behav. 9, 141 (1985); A. W. Martin and D. A. Schum,
judgmental thoroughness and accuracy, legal competence, impartial- JurimetricsJ. 383 (Summer 1987); R. J. Simon and L. Mahan, Law Soc. Rev. 5, 319
ity, representativeness, consistency, efficiency, and perceived legiti- (1971).
22. R. J. MacCoun, Dissertation Abstr. Int. 46, 700B (1984).
macy-only some of which can be assessed by mock jury research
23. J. Kaplan, Stanford Law Rev. 20, 1065 (1968); S. S. Nagel and M. G. Neef, Decision
(40). Although it can be readily shown that jury performance falls Theory in the Legal Process (Lexington Books, Lexington, MA, 1979); L. H. Tribe,
short of ideal standards on some of these dimensions, the critical Harvard Law Rev. 84, 1329 (1971).
24. W. L. Bennett and M. S. Feldman, Reconstructing Reality in the Courtroom (University
questions for public policy are (i) under what conditions can jury Publications of America, Frederick, MD, 1984); N. Pennington and R. Hastie, J.
performance be enhanced, and (ii) how does the jury perform Pers. Soc. Psychol. 51, 242 (1986).
relative to other legal decision-makers? 25. R. C. Dillehay and M. T. Nietzel, in The Impact of Social Psychology on Procedural
Justice, M. F. Kaplan, Ed. (Thomas, Springfield, IL, 1986), p. 167; R. S. Tindale
Researchers have begun to address the first question, but less is and D. H. Nagao, Org. Behav. Hum. Decision Proc. 37, 409 (1986).
known about the second one. In a survey of more than 3500 26. S. Penrod and D. Linz, in The Impact of Social Psychology on ProceduralJustice, M. F.
Kaplan, Ed. (Thomas, Springfield, IL, 1986), p. 135. Research indicates that
criminal jury trials conducted in the 1950s (31), the judge agreed
individual-difference variables are more predictive of broad patterns of behavior
with the jury's verdict more than 75% of the time, but the sources of than of single behaviors like voting for a verdict [M. Snyder and W. Ickes, in The
the judge-jury disagreements are still not understood. It is not clear Handbook of Social Psychology, G. Lindzey and E. Aronson, Eds. (Rand McNally,
New York, ed. 3, 1985), vol. 2, p. 883].
whether these disagreements indicate that judges and jurors evaluate
27. See the Amicus Curiae brief submitted by the American Psychological Association
testimony differently or apply different standards of proof, in
orLockhart v. McCree, 106 S. Ct. 1758 (1986), reprinted in Am. Psychol. 42, 59
whether judges and juries follow completely different judgment (1987).
28. Reviewed by F. C. Dane and L. S. Wrightsman, in The Psychology of the Courtroom,
processes. Experimental trial simulations that compare the processes N. L. Kerr and R. M. Bray, Eds. (Academic Press, New York, 1982), p. 83; J.
by which juries, professional trial judges, and other legal fact finders Carroll et al., Law Hum. Behav. 10, 187 (1986).
29. T. R. Carretta and R. L. Moreland, J. Appl. Soc. Psychol. 13, 291 (1983).
reach their verdicts may provide answers to this question.
30. N. L. Kerr, Pers. Soc. Psychol. Bull. 4, 479 (1978); J. Pers. Soc. Psychol. 36, 1431
(1978); , R. Bull, R. J. MacCoun, H. Rathborn, Brit. J. Soc. Psychol. 24,47
(1985).
31. H. Kalven and H. Zeisel, The AmericanJury (Little, Brown, Boston, 1966); M. H.
REFERENCES AND NOTES Walsh, Yale LawJ. 79, 142 (1969); A. E. Bottoms and M. A. Walker, J. Am. Stat.
Assoc. 67, 773 (1972).
1. There is an extensive literature on the history and behavior of juries; see, for 32. J. H. Davis, in Progress in Social Psychology, M. Fishbein, Ed. (Erlbaum, Hillsdale,
example, P. Hans and N. Vidmar, Judging theJury (Plenum, New York, 1986). NJ, 1980), p. 157; D. A. Vollrath and J. H. Davis, in TheJury: Its Role in American
2. For example, in a series of opinions in the 1970s, the Supreme Court asserted that Society, R. J. Simon, Ed. (Lexington Books, Lexington, MA, 1980); B. Grofman,
reductions in the jury's size (Williams v. Florida, 399 U.S. 78, 1970; Colgrove v. in The Trial Process, B. D. Sales, Ed. (Plenum, New York, 1981), p. 305; S. Penrod
Battin, 413 U.S. 149, 1973; Ballew v. Georgia, 435 U.S. 222, 1978) and a and R. Hastie, Psychol. Bull. 86, 462 (1979); G. Stasser, N. L. Kerr, R. M. Bray, in
relaxation of the traditional unanimity requirement (Apodaca, Cooper and Madden v. Psychology of the Courtroom, N. L. Kerr and R. M. Bray, Eds. (Academic Press,
The
Oregon, 406 U.S. 404, 1970; Johnson v. Louisiana, 406 U.S. 356, 1972) would New York, 1982), p. 221.
have minimal effects on jury representativeness, performance, and verdicts. 33. 5. D. Penrod and R. Hastie, Psychol. Rev. 87, 133 ( 1980) with erratum on p. 476;
3. R. J. MacCoun, Getting Inside the Black Box: Toward a Better Understanding of Civil G. Stasser and J. H. Davis, ibid. 88, 523 (1981).
Jury Behavior (RAND, Santa Monica, CA, 1987). 34. D. J. Isenberg, J. Pers. Soc. Psychol. 50, 1141 (1986); D. G. Myers and H. Lamm,
4. E. Campbell, Monash Univ. Law Rev. 11, 169 (1985). Psychol. Bull. 83, 602 (1976).
5. D. R. Hensler, M. E. Vaiana, J. S. Kakalik, M. A. Peterson, Trends in Tort Litigation:
35. N. L. Kerr and R. J. MacCoun, J. Pers. Soc. Psychol. 48, 349 (1985).
The Story Behind the Statistics (RAND, Santa Monica, CA, 1987); M. A. Myers, 36. LawN. L. Kerr, ibid. 41, 684 (1981); , R. J. MacCoun, C. H. Hansen, J. A.
Soc. Rev. 13, 781 (1979); C. M. Werner, M. J. Strube, A. M. Cole, D. K. Hymes, J. Exp. Social Psychol. 23, 119 (1987).

2 JUNE i989 ARTICLES 1049

This content downloaded from 49.150.13.208 on Thu, 07 Feb 2019 03:49:05 UTC
All use subject to https://about.jstor.org/terms
37. M. F. Kaplan and C. E. Miller, J. Pers. Soc. Psychol. 53, 306 (1987); S. Tanford and W. D. Lindblom, J. Appl. Soc. Psychol. 16, 686 (1986).
and S. Penrod, J. Appl. Soc. Psychol. 16, 322 (1986). 42. J. C. Mowen and D. E. Linder, in Social Psychology and Discretionary Law, L. E. Abt
38. J. Hartwick, B. H. Sheppard, J. H. Davis, in Improving Group Decision Making in and I. R. Stuart, Eds. (Van Nostrand Reinhold, New York, 1979), p. 219; N.
Organizations, R. Guzzo, Ed. (Academic Press, New York, 1982), p. 41. Vidmar, J. Pers. Soc. Psychol., 22, 211 (1972).
39. J. H. Davis, C. E. Spitzer, D. H. Nagao, G. Stasser, in Dynamics of Group Decisions, 43. K. S. Bordens and I. A. Horowitz, Law Hum. Behav. 9, 339 (1985); S. Tanford, S.
H. Brandstatter, J. H. Davis, H. Schuler, Eds. (Sage, Beverly Hills, CA 1978); V. Penrod, R. Collins, ibid., p. 319.
P. Hans and A. N. Doob, Criminal Law 18, 235 (1976). 44. Supported by NSF grant SES-8796357 and by The RAND Corporation's
40. Citizens' evaluations of the desirability of judges and different types of juries are Behavioral Science Department. I am grateful to L. Dair, D. Hensler, D. Kanouse,
examined in R. J. MacCoun and T. R. Tyler, Law Hum. Behav. 12, 333 (1988). N. Kerr, B. Mittman, J. Peterson, J. Rolph, and T. Tyler for helpful comments on
41. C. F. Roberts, S. L. Golding, F. D. Fincham, ibid. 11, 207 (1987); J. C. Savitsky an earlier draft.

The Heliosphere as an Astrophysical Laboratory

for Particle Acceleration
T. TERASAWA AND M. SCHOLER

able to make in situ observations of shock waves from supernovae.

Particle acceleration is one of the most important topics Because physics is an experimental science, we need a laboratory in
in plasma astrophysics as well as in cosmic-ray astrophys- which we can test basic principles. In this respect, the heliosphere,
ics. The heliosphere is an ideal astrophysical laboratory, within which the solar wind plasma has the greatest influence on the
wherein one can observe in situ the elementary mecha- dynamics and the energetics of particles, can be considered as a
nisms involved in the particle acceleration processes. Two laboratory for particle acceleration processes. Of course, the parame-
phenomena of special interest are stochastic acceleration ter regime that we can reach in heliospheric observations is rather
in the magnetohydrodynamic turbulence around comets limited. Nevertheless, in this way we can learn much about the
and stochastic shock acceleration at interplanetary shock elementary principles that govern particle acceleration processes.
waves. Let us start with an elementary consideration of the motion of
charged particles in cosmic plasmas permeated by magnetic fields.
Charged particles are most efficiently accelerated by the electric field,
and the effects of other forces, such as gravity, are negligible.
JN DILUTE PLASMAS IN THE UNIVERSE, THE USUAL IWO-BODY However, the electric field does not always accelerate these particles
Coulomb collisions are relatively unimportantfreely
and ifthethebehavior
background plasma is steady and homogeneous: EI,
the electric field component perpendicular to the magnetic field, B,
of charged particles is governed by collective interactions
through long-range electromagnetic forces. When some dynamical can only produce EI x B drift motion (that is, magnetized mo-
energy release occurs in these plasmas, a part of the thermal tion). Suppose that the magnetic field is directed toward the page
population is accelerated to high energies, so that the particle and that the electric field is directed downward (Fig. 1). A particle
distribution deviates significantly from the Maxwell-Boltzmann (of mass m), which is initially at rest, makes a cycloid motion in
distribution. By studying the particle acceleration process, we can configuration space (Fig. la). The average speed of this particle is
understand the detailed physics of the energy conversion process E xB (1)
and recognize what extreme conditions are attained by particles in VE=C IBI
the system.
The latest example from astrophysical observations is the superno- where c is the vacuum speed of light. In velocity space, the orbit of
va explosion that occurred early in 1987 [designated SN1987A (1)]. this particle is a circle (Fig. Ib). The maximum velocity a particle can
Electromagnetic radiation, from radio waves to gamma rays, as wellobtain is 21VE1. The corresponding maximum energy, 2mlVE12, is
as a neutrino burst have been detected. Astrophysicists are anxious now known as the maximum "pickup" energy. (This name is derived
from the study of cometary ions. See the next section.) To accelerate
to detect new signals, which are either directly or indirectly related
to the particle acceleration processes, such as acceleration at the this particle to energies higher than this maximum pickup energy, it
supernova shock wave, stochastic acceleration in the turbulence is necessary to break the magnetized motion so that the particle can
generated in the supernova ejecta, or acceleration by the strong move along the direction of E1 . In the following sections, we shall
pulsar electric field. Astrophysical observations, however, are re- discuss stochastic acceleration processes in unsteady plasmas, in
mote-sensing observations. In the foreseeable future, we will not bewhich such a breakdown becomes possible.
Another way to accelerate particles efficiently is to have an electric
field component parallel to the magnetic field, E 11. Because the
T. Terasawa is in the Department of Geophysics, Kyoto University, Kyoto 606, Japan.
magnetic field does not prevent particle acceleration along it, E
M. Scholer is at the Institut fur extraterrestrische Physik, Max-Planck-Institut ffir Physik
und Astrophysik, 8046 Garching, Federal Republic of Germany. can accelerate particles freely. However, E 11 is easily short-circuited

1050 SCIENCE, VOL. 244

This content downloaded from 49.150.13.208 on Thu, 07 Feb 2019 03:49:05 UTC
All use subject to https://about.jstor.org/terms