ArticlePDF Available

‘They Saw a Protest’: Cognitive Illiberalism and the Speech-Conduct Distinction


Abstract and Figures

“Cultural cognition” refers to the unconscious influence of individuals’ group commitments on their perceptions of legally consequential facts. We conducted an experiment to assess the impact of cultural cognition on perceptions of facts relevant to distinguishing constitutionally protected “speech” from unprotected “conduct.” Study subjects viewed a video of a political demonstration. Half the subjects believed that the demonstrators were protesting abortion outside of an abortion clinic, and the other half that the demonstrators were protesting the military’s “don’t ask, don’t tell” policy outside a campus recruitment facility. Subjects of opposing cultural outlooks who were assigned to the same experimental condition (and who thus had the same belief about the nature of the protest) disagreed sharply on key “facts” – including whether the protesters obstructed and threatened pedestrians. Subjects also disagreed sharply with those who shared their cultural outlooks but who were assigned to the opposing experimental condition (and hence had a different belief about the nature of the protest). These results supported the study hypotheses about how cultural cognition would affect perceptions pertinent to the “speech”-“conduct” distinction. We discuss the significance of the results for constitutional law and liberal principles of self-governance generally.
Content may be subject to copyright.
Electronic copy available at:
“They Saw a Protest”: Cognitive Illiberalism and the
Speech-Conduct Distinction
Dan M. Kahan,
David A. Hoffman,
Donald Braman,
Danieli Evans,
& Jeffrey J. Rachlinski
Cultural cognition refers to the unconscious influence of individuals’ group
commitments on their perceptions of legally consequential facts. We con-
ducted an experiment to assess the impact of cultural cognition on percep-
tions of facts relevant to distinguishing constitutionally protected “speech”
from unprotected “conduct.” Study subjects viewed a video of a political
demonstration. Half the subjects believed that the demonstrators were pro-
testing abortion outside of an abortion clinic, and the other half that the de-
monstrators were protesting the military’s “don’t ask, don’t tell” policy out-
side a campus recruitment facility. Subjects of opposing cultural outlooks
who were assigned to the same experimental condition (and thus had the
same belief about the nature of the protest) disagreed sharply on key
“factsincluding whether the protestors obstructed and threatened pede-
strians. Subjects also disagreed sharply with those who shared their cultural
outlooks but who were assigned to the opposing experimental condition (and
hence had a different belief about the nature of the protest). These results
supported the study hypotheses about how cultural cognition would affect
perceptions pertinent to the “speech”-“conduct” distinction. We discuss the
significance of the results for constitutional law and liberal principles of self-
governance generally.
I. Introduction ............................................................................................................. 3
II. Theoretical Background ......................................................................................... 4
A. “Speech” vs. “Conduct”................................................................................... 5
B. Culturally Motivated Reasoning ..................................................................... 8
Yale Law School.
Temple University Beasley School of Law.
George Washington University Law School.
Cultural Cognition Project Lab at Yale Law School.
Cornell Law School.
The study featured in this paper was funded by the Oscar M. Ruebhausen Fund at Yale Law School,
Temple University Beasley School of Law, and George Washington University Law School. We are
grateful to members of the Harry Phillips American Inn of Court in Nashville, Tenn., for their ge-
nerous participation in, and thoughtful feedback on, a pretest conducted to assess the study design.
We are also grateful to Kw Bilz, Jeffrey Dunoff, Bill Eskridge, Janice Nadler, Richard Posner, David
Sherman, Dan Simon, and Avani Sood for comments.
64 Stan. L. Rev. (2012)
subject to revision
Electronic copy available at:
“They Saw a Protest” 2
C. Cognitive Illiberalism and the Speech-conduct Distinction .......................... 10
III. Study ...................................................................................................................... 11
A. Overview and Hypotheses ............................................................................... 11
B. Design and Methods ........................................................................................ 16
1. Sample ......................................................................................................... 16
2. Cultural Worldviews ................................................................................... 17
3. Stimulus ....................................................................................................... 17
4. Response Measures ..................................................................................... 22
5. Analytic Strategy .......................................................................................... 23
C. Results .............................................................................................................. 23
IV. Analyzing, Appraising, and Advocating ............................................................... 28
A. Summary of Results ......................................................................................... 29
B. Cognitive Illiberalism and the Constitution ................................................... 31
C. Judges, Jurors, and Citizens............................................................................. 34
D. Debiasing ......................................................................................................... 40
1. Affirmation and Jury Selection ................................................................... 40
2. Deliberative Depolarization ........................................................................ 41
3. Judicial Aporia .............................................................................................. 43
V. Conclusion ............................................................................................................. 44
Appendix. Study Instrument ...................................................................................... 48
Anyone seriously interested in what this case was about
must view that tape. And anyone doing so who is familiar
with run-of-the-mine labor picketing, not to mention some
other social protests, will be aghast at what it shows we
have today permitted an individual judge to do.
Madsen v. Women’s Health Center, 512 U.S.
753, 786 (1994) (Scalia, J., dissenting).
Justice Stevens suggests that our reaction to the videotape
is somehow idiosyncratic, and seems to believe we are mi-
srepresenting its contents. See post, at 1783 (dissenting
opinion) (“In sum, the factual statements by the Court of
Appeals quoted by the Court ... were entirely accurate”).
We are happy to allow the videotape to speak for itself. See
Record 36, Exh. A, available at http:// www. supremecour-
tus. gov/ opinions/ video/ scott_ v_ harris. rmvb and in
Clerk of Court's case file.
Scott v. Harris, 550 U.S. 372, 378 n.5 (2007)
(Scalia, J.)
Electronic copy available at:
“They Saw a Protest” 3
In a 1950s social psychology experiment, students from two Ivy League col-
leges were instructed to evaluate a series of controversial officiating calls made
during a football game between their respective schools. Researchers found that
the students, from both institutions, were much more likely to perceive error in
the penalty assessments imposed on their school’s team than in those imposed on
their rival’s. The students’ emotional stake in affirming their loyalty to their institu-
tions, researchers concluded, had unconsciously shaped what they had seen when
viewing events captured on film.
This study is now recognized as a classic dem-
onstration of motivated cognition, the ubiquitous tendency of people to form percep-
tions, and to process factual information generally, in a manner congenial to their
values and desires.
Motivated cognition poses an obvious hazard for law. Sports fans are permit-
tedeven expectedto be partisan. But legal decisionmakers must be neutral.
Just as the integrity of a sporting contest would be undermined by unconscious
favoritism on the part of the referee, so the legitimacy of the law would likewise be
compromised if legal decisionmakers, as a result of motivated cognition, unwit-
tingly formed perceptions of facts that promoted the interests and values of
groups with whom they had an affinity.
This effect could be particularly subversive of constitutional law. The Free
Speech, Equal Protection, and Due Process Clauses all mandate governmental
even-handedness. Within their respective domains, each forecloses the State from
privileging particular affiliations, ways of life, or points of view and mandates that
law be justified by its contribution to secular interestsphysical security, public
health, economic prosperityvalued by all citizens.
But if decisionmakers (par-
See Albert H. Hastorf & Hadley Cantril, They Saw a Game: A Case Study, 49 J. Abnormal & Social
Psychol. 129 (1954).
See generally Ziva Kunda, The Case for Motivated Reasoning, 108 Psychol. Bull. (1990); Roger Giner-
Sorolla & Shelly Chaiken, Selective Use of Heuristic and Systematic Processing Under Defense Motivation, 23
Personality & Social Psychol. Bull. 84 (1997); Emily Balcetis & David Dunning, See What You Want to
See: Motivational Influences on Visual Perception, 91 J. Personality & Social Psychol. 612 (2006); Anca M.
Miron, Nyla R. Branscombe & Monica Biernat, Motivated Shifting of Justice Standards, 36 Personality &
Social Psychol. Bull. 768 (2010).
See Dan M. Kahan & Donald Braman,
The Self-defensive Cognition of Self-defense, 45 Am. Crim. L. Rev.
1 (2008). A number of recent studies examine motivated cognition in law. See Avani M. Sood & John
M. Darley, The Plasticity of Harm in the Service of Punishment Goals: An Experimental Demonstration, Calif. L.
Rev. (forthcoming 2012), paper. Available at SSRN:; Janice Nad-
ler, The Psychology of Blame: Criminal Liability and the Role of Moral Character, 97 Cornell L. Rev. (forth-
coming 2012);
Dan M. Kahan, Culture, Cognition, and Consent: Who Perceives What, and Why, in Acquain-
tance Rape Cases, 158 Univ. Pa. L. Rev. 729 (2010); Dan M. Kahan, David A. Hoffman & Donald
Braman, Whose Eyes are You Going to Believe? Scott v. Harris and the Perils of Cognitive Illiberalism, 122
Harv. L. Rev. 837-906 (2009). See generally Dan M. Kahan,
The Supreme Court 2010 TermForeword:
Neutral Principles, Motivated Cognition, and Some Problems for Constitutional Law, 126 Harv. L. Rev. 1, 59-66
See generally Ronald Dworkin, Freedom’s Law: The Moral Reading of the American Constitution
(1999); David A. J. Richards, Toleration and the Constitution (1986).
“They Saw a Protest” 4
ticularly adjudicators) unconsciously apply these provisions to favor outcomes
congenial to favored ways of life, citizens who adhere to disfavored ones will suf-
fer the same array of disadvantages for failing to conform that they would in a re-
gime expressly dedicated to propagation of a sectarian orthodoxy. This distinctive-
ly psychological threat to constitutional ideals, which we will refer to as “cognitive
has received relatively little attention from commentators or jurists.
We performed an experimental study designed to help assess just how much
of a threat cognitive illiberalism poses to constitutional ideals. The study focused
on a discrete and recurring task in constitutional law: discernment of the line be-
tween “speech” and “conduct” for purposes of the First Amendment. Embodied
in a variety of doctrines, the speech-conduct distinction aims to assure that coer-
cive regulation is justified on grounds unrelated to governmental or public hostility
to disfavored ideas.
Most importantly, the “speech”-“conduct” distinction has
historically played, and continues to play, a vital function in preventing the gov-
ernment from invoking its responsibility for maintaining “public order” to disguise
suppression of impassioned political dissent.
Our study furnishes strong evidence
that this function is indeed highly vulnerable to the power of motivated cognition
to shape decisionmakers’ perceptions of the facts that mark the speech-conduct
The features of the “speech-conduct” distinction that make it susceptible to
this influence, moreover, are shared by a host of other constitutional doctrines.
The study results thus highlight the need to fortify constitutional theorizing with
psychological realism. Normatively ideal standards for enforcing the Constitution
are of little value if applying them defies the capacities of constitutional decision-
Following background discussion, we describe the study design and results.
Thereafter, we address the study’s normative and prescriptive implications.
The context for our study comprises three elements. The first is the
“speech”-“conduct” distinction in First Amendment doctrine. The second is the
phenomenon of culturally motivated cognition. And the third is the threat the latter
poses to the former.
See generally Dan M. Kahan,
The Cognitively Illiberal State, 60 Stan. L. Rev. 115 (2007).
For a provocative and insightful exception, see Sood & Darley, supra note 3.
See generally Elena Kagan, Private Speech, Public Purpose: The Role of Governmental Motive in First Amend-
ment Doctrine, 63 U. Chi. L. Rev. 413, 428-29 (1996).
Terminello v. Chicago, 337 U.S. 1 , 4 (1949) (it is “a function of free speech under our system of gov-
ernment is to invite dispute. It may indeed best serve its high purpose when it induces a condition of
unrest, creates dissatisfaction with conditions as they are, or even stirs people to anger. Speech is
often provocative and challenging”); Texas v. Johnson, 491 U.S. 397, 408-9 (1989) (First Amendment
does not permit speech to be restricted on ground that “an audience that takes serious offense at
particular expression is necessarily likely to disturb the peace”).
“They Saw a Protest” 5
A. “Speech” vs. “Conduct”
Because the Free Speech Clause confers special protection on speech, First
Amendment jurisprudence is said to “draw vital distinctions between words and
deeds, . . . ideas and conduct.”
These sorts of divisions, however, are notoriously
Words are often the keysometimes the exclusiveinstruments of
prohibited forms of conduct, from price-fixing
to treason.
Deeds such as lighting
fire to an American flag or to a towering crossnot to mention violently assault-
ing a person on account of his race or sexual preferencecan potently express
ideas. In short, we “do things with words and say things with actions.”
that every act be definitively categorized as either speech or conducta position
John Hart Ely called the “ontological fallacy”thus invites sophism and ad ho-
One way to avoid this problem is to adopt instead what Ely referred to as a
“teleological” conception of the speech-conduct distinction.
Rather than direct-
ing courts to determine whether a particular act is “really” expression or “really”
conduct, this approach focuses attention on the government’s goal in regulating it.
The “bedrock principle underlying the First Amendment . . . is that the govern-
ment may not prohibit the expression of an idea simply because society finds the
idea itself offensive or disagreeable.”
It therefore makes sense to treat a regula-
tion as abridging speech whenever the government’s purpose is to attain some good
or state of affairs that reflects aversion to a disfavored idea.
If, in contrast, a reg-
Ashcroft v. Free Speech Coalition, 535 U.S. 234, 253 (2002).
See, e.g., Barnes v. Glen Theatre, 501 U.S. 560, 576 (1991) (Scalia, J., concurring) (“Virtually every law
restricts conduct, and virtually any prohibited conduct can be performed for an expressive purpose-if
only expressive of the fact that the actor disagrees with the prohibition.”); John Hart Ely, Flag Dese-
cration: A Case Study in the Roles of Categorization and Balancing in First Amendment Analysis, 88
Harv. L. Rev. 1482, 1495 (1975) (“Burning a draft card to express one’s opposition to the draft is an
undifferentiated whole, 100% action and 100% expression, and to outlaw the act is therefore neces-
sarily to regulate both elements.”); Louis Henkin, The Supreme Court, 1967 Term Foreword: On Draw-
ing Lines, 82 Harv. L. Rev. 63, 77-80 (1968) (arguing that the “distinction between speech and non-
speech has no content” and is “specious”). Thomas Emerson is the constitutional theorist most
famously associated with the distinction. See Thomas I. Emerson, The System of Freedom of Ex-
pression (1970).
Cf. F.T.C. v. Superior Court Trial Lawyers Ass'n, 493 U.S. 411, 427 (1990) (boycott conducted to
effect increase in prices not protected by First Amendment).
See R.A.V. v. St. Paul, 505 U.S. 377, 389 (1992) (“[W]ords can in some circumstances violate laws
directed not against speech but against conduct (a law against treason, for example, is violated by
telling the enemy the Nation’s defense secrets).”)
Jed Rubenfeld, The First Amendment's Purpose, 53 Stan. L. Rev. 767, 783-84 (2001); Henkin, supra
note 10, at 79 (“Speech is conduct, and actions speak.”).
See Ely, supra note 10, 1494-95.
See id. at 1496.
Texas v. Johnson, 491 U.S. 397, 414 (1989).
See Ely, supra note 10, at 1496-1500; Kagan, supra note 7, at 428-32; Rubenfeld, supra 13, at 777.
“They Saw a Protest” 6
ulation seeks to promote a good that can be defined independently of hostility to a
disfavored idea, we can say that a violator, even if she intends to communicate a
message, is being punished for engaging in “illegal conduct,” not “for speaking.”
The Supreme Court has used the teleological strategy to distinguish “speech”
from “conduct” across a diverse range of settings. The government can ban sleep-
ing overnight in Lafayette Park to protest homelessness, for example, not because
sleeping just can’t be “speech,” but because the government’s reason for the ban is
“unrelated to suppression of free expression”: “limit[ing] wear and tear on park
properties” justifies prohibiting overnight camping there regardless of whether the
campers mean to express a message or what it might be.
The government can criminalize the burning of draft cards,
the Court has
held, but not the burning of American flags.
The basis for the distinction isn’t
that the latter is more “speech like” than the former; indeed, both might be recog-
nized (and were in the 1960s) as cogent statements of opposition to a war. The
difference stems from the government’s reasons for regulating them. Preserving
ready proof of compliance with selective service laws supplies a justification for
prohibiting destruction of draft cards independent of any hostility toward the
statement of dissent such behavior might express; accordingly, the government’s
interest in prohibiting the burning of them is (once more) “unrelated to the sup-
pression of free expression.”
The government’s interest in banning the burning
of American flags, however, is not. “Preserving the flag as a symbol of nationhood
and national unity” necessarily involves favoring one set of messages over anoth-
Nor can “preventing breaches of the peace” be viewed as a justification in-
dependent of hostility toward a disfavored message if the only cause for such dis-
order is the “serious offen[se]” onlookers would take toward the burning of the
The government’s interest in protecting individuals from “distinct emotional
harms” and in averting retaliatory cycles of violence supplies “an adequate expla-
nation” for “hate crime” laws “over and above mere disagreement with offenders’
beliefs or biases,”
the Court has reasoned. Likewise, protecting individuals from
fear of physical attack is a constitutionally sound basis for prohibiting dramatic ges-
Rubenfeld, supra note 13, at 778.
Clark v. CCNV, 468 U.S. 288, 295, 299 (1984).
United States v. O’Brien, 391 U.S. 367 (1968).
Texas v. Johnson, 491 U.S. 397 (1989); United States v Eichman, 496 U.S. 310 (1990).
O’Brien, 391 U.S. at 377.
Johnson, 491 U.S. at 407; see also Eichman, 496 U.S. at 316-17 (protection of meaning of flag as
“symbol of national unity” cannot be understood without reference to interest in regulating the ideas
associated with various uses of flag).
Johnson, 491 U.S. at 408-09.
Wisconsin v. Mitchell, 508 U.S. 476, 488 (1993).
“They Saw a Protest” 7
tures, such as cross-burnings, intended to intimidate.
Nevertheless, if the selec-
tivity with which the government prohibits such assaultive behavior reflects a
“special hostility towards the particular biases thus singled out,” punishment of
such conduct reflects exactly the sort of “disapproval of ideas” that the First
Amendment is meant to proscribe.
In addition to systematizing a diverse body of cases, the teleological concep-
tion of the speech-conduct distinction also integrates First Amendment doctrine
into a more general theory of constitutional liberty. The prohibition on state en-
dorsement of a partisan conception of the good lifethe core tenet of liberal neu-
is reflected in the First Amendment injunction that “no official, high or
petty, can prescribe what shall be orthodox in politics, nationalism, religion, or
other matters of opinion.”
The principle of “liberal public reason,” which re-
quires that law be justified by its contribution to attainment of secular goods of
value to citizens of diverse cultural and moral outlooks,
is advanced when courts
scrutinize the asserted basis of regulations to assure that they advance interests
“unrelated to suppression” of disfavored ideas.
Decisions construing equal pro-
and due process
to forbid imposition of other types of legal disabilities
See Virginia v. Black, 538 U.S. 343 (2003).
R.A.V. v. St. Paul, 505 U.S. 377, 396 (1992).
See, e.g., Ronald Dworkin, Liberalism, in Liberalism and Its Critics 60, 63-64 (M. Sandel ed., 1984).
W. Va. Bd. of Educ. v. Barnette, 319 U.S. 624, 642 (1943); see Dworkin, supra note 4, at 237-38 (First
Amendment reflects liberal principle that “no one may be prevented from influencing the shared
moral environment, through his own private choices, tastes, opinions, and example, just because
these tastes or opinions disgust those who have the power to shut him up or lock him up”); Note, A
Communitarian Defense of Group Libel Laws, 101 Harv. L. Rev. 682, 688 (1988) (arguing that First
Amendment implements bar on state endorsement of the good by treating aversion that some per-
sons feel toward the life choices of others” as a noncognizable harm).
See John Rawls, Political Liberalism 175, 217-18 (1993) (articulating norm of “public reason” that
prohibits political actors in most contexts from invoking “comprehensive views” that “in-
clude[]conceptions of what is of value in human life, as well as ideals of personal virtue and charac-
ter” and instead “explain . . . how the principles and policies they advocate and vote for can be sup-
ported by” considerations consistent with “a diversity of reasonable religious and philosophical doc-
trines”); see also David A. Strauss, Legal Argument and the Overlapping Consensus 20-21 (unpub-
lished, July 12, 1998) (arguing that conventional modes of legal reasoning and justification reflect a
liberal public-reason norm).
See, e.g., Kagan, supra note 7, at 453-54 (“the strict scrutiny standard . . . is best understood as an
evidentiary device” to furnish “assurance that the government has acted for proper reasons” and that
the “interested asserted” is not a “pretext” for “antipathy for the speech affected” by regulation).
See Romer v. Evans, 517 U.S. 620, 632, 633 (1996) (“ ‘[I]f the constitutional conception of ‘equal
protection’ of the laws' means anything, it must at the very least mean that a bare . . . desire to harm a
politically unpopular group cannot constitute a legitimate governmental interest. ”) quoting (Depart-
ment of Agriculture v. Moreno, 413 U.S. 528, 534 (1973)).
See Lawrence v. Texas, 539 U.S. 558, 567-71 (2003) (Due Process Clause forbids the “majority [to]
use the power of the State to enforce . . . on the whole society” standards of “private conduct” that
originate in “religious beliefs, conceptions of right and acceptable behavior, and respect for the tradi-
tional family”).
“They Saw a Protest” 8
solely to promote favored moral and religious norms can be read in like fashion.
Distinguishing “speech” from “conduct,” then, can be seen as characteristic of the
type of judgments courts must make to perfect the liberal underpinnings of the
American constitutional regime.
B. Culturally Motivated Reasoning
Cultural cognition is a species of motivated reasoning that promotes congru-
ence between a person’s defining group commitments, on the one hand, and his
or her perceptions of risk and related facts, on the other.
A variety of mechan-
isms contribute to this effect. Thus, individuals tend selectively to credit empirical
information in patterns congenial to their cultural values.
They are also disposed
to impute knowledge and expertise to others with whom they share a cultural af-
And they are more likely to note, assign significance to, and recall facts
supportive of their cultural outlooks than facts subversive of them.
These dy-
namics protect individuals’ connection to others on whom they depend for ma-
terial and emotional support.
At a societal level, however, culturally motivated cognition can be a source of
intense and enduring political conflict.
Citizens who subscribe to an egalitarian
ethic that identifies free markets as fonts of unjust disparity readily credit evidence
that commerce and industry are destroying the environment; citizens who adhere
See generally Dworkin, supra note 4, at 110-12 (arguing that right of individuals “to confront for
themselves, answering to their own consciences and convictions, the most fundamental questions
touching the meaning and value of their own lives” inheres in “the structure of the Constitution” as
well as in various textual provisions of it); Dworkin, supra note 28, at 70 (asserting that “the rights
encoded in the Bill of Rights of the United States Constitution, as interpreted (on the whole) by the
Supreme Court, are those that a substantial number of liberals would think reasonably well suited to
what the United States now requires . . . .”); Richards, supra note 4 (using liberal theory to explicate
constitutional guarantees of free speech, freedom of religion, equality, and privacy).
Cf. Kagan, supra note 7, at 511 (suggesting that the First Amendment prohibition on making aver-
sion to ideas a basis for regulating reflects the “principle that the government must treat all persons
with equal respect and concern” and that the same “principle may well explain much of equal protec-
tion law”).
For this reason, we refer interchangeably to “cultural cognition” and “culturally motivated cogni-
tion” or “culturally motivated reasoning.”
See, e.g., Dan M. Kahan, Donald Braman, Paul Slovic, John Gastil & Geoff Cohen,
Cultural Cogni-
tion of the Risks and Benefits of Nanotechnology, 4 Nature Nanotechnology 87 (2009).
See, e.g., Dan M. Kahan, Donald Braman, Geoffrey L. Cohen, John Gastil & Paul Slovic,
Fears the HPV Vaccine, Who Doesn't, and Why? An Experimental Study of the Mechanisms of Cultural Cogni-
tion, 34 L. & Human Behavior 501 (2010).
See generally Dan M. Kahan, Hank Jenkins-Smith & Donald Braman,
Cultural Cognition of Scientific
Consensus, 14 J. Risk Res. 147 (2011).
See generally David K. Sherman, Geoffrey L. Cohen & P. Zanna Mark, The Psychology of Self-defense:
Self-Affirmation Theory, in 38 Advances in Experimental Social Psychology 183 (2006).
See Dan Kahan,
Fixing the Communications Failure, 463 Nature 296-297 (2010).
“They Saw a Protest” 9
to an individualistic ethic that prizes private orderings dismiss such evidence and
insist instead that needless government regulation threatens to wreck economic
Associating firearms with patriarchy, racism, and distrust, egalitarian
and communitarian citizens blame accidental shootings and crime on insufficient
regulation of guns; hierarchical and individualist citizens, in contrast, worry that too
much regulation will render law-abiding citizens vulnerable to predation, a belief
congenial to the value they attach to guns as instruments of social roles (father,
protector) and symbols of virtues (self-reliance, honor) distinctive of their ways of
Citizens who combine hierarchical and communitarian values believe that
the right to abortion demeans those women who eschew the workplace to be
mothers; correspondingly, they worry that abortion poses a health risk to women.
Citizens who combine egalitarian and individualist values, and who assign status to
women as well as men for professional and commercial success, believe that restric-
tions on abortion put women’s health in danger.
Myriad other issuesfrom the
risks and benefits of the HPV vaccine for schoolgirls
to the efficacy of legally
mandated medical treatment for (noninstitutionalized) mentally ill persons
divide citizens along lines that correspond to the social meanings these policies
connote within opposing ways of life.
Conflicts of this sort expose democratic pluralism to a distinctive threat: cogni-
tive illiberalism. Because their perceptions of risk and related facts are unconscious-
ly motivated by their defining commitments, even citizens who are genuinely
committed to principles of liberal neutrality are likely to end up persistently di-
vided along cultural linesnot over the proper ends of law (physical security, eco-
nomic prosperity, public health, and the like) but over the means for securing them.
Nor is the cultural complexion of these seemingly empirical disputes likely to
evade notice by those involved in them. On the contrary, consistent with a dynam-
ic known as “naïve realism,” each side in these conflicts is likely to suspect the
See Mary Douglas & Aaron B. Wildavsky, Risk and Culture: An Essay on the Selection of Tech-
nical and Environmental Dangers (1982); Karl Dake, Orienting Dispositions in the Perception of Risk: An
Analysis of Contemporary Worldviews and Cultural Biases, 22 J. Cross-Cultural Psych. 61 (1991); Dan M.
Kahan & Donald Braman, Cultural Cognition of Public Policy, 24 Yale J. L. & Pub. Pol’y 147 (2006).
Dan M. Kahan, Donald Braman, John Gastil, Paul Slovic & C. K. Mertz,
Culture and Identity-
Protective Cognition: Explaining the White-Male Effect in Risk Perception, 4 Journal of Empirical Legal Stu-
dies 465 (2007).
See id.
See Dan M. Kahan, Donald Braman, Geoffrey L. Cohen, John Gastil & Paul Slovic,
Who Fears the
HPV Vaccine, Who Doesn't, and Why? An Experimental Study of the Mechanisms of Cultural Cognition, 34 L.
& Human Behavior 501 (2010).
See Dan M. Kahan, Donald Braman, John Monahan, Lisa Callahan & Ellen Peters, Cultural Cogni-
tion and Public Policy: The Case of Outpatient Commitment Laws, 34 L. & Human Behavior 118 (2010).
See R. Gutierrez & R. Giner-Sorolla, Anger, Disgust, and Presumption of Harm as Reactions to Taboo-
breaking Behaviors, 7 Emotion (2007) (finding that individuals are motivated to impute harm to intrin-
sically immoral behavior); Jonathan Haidt & Mathew. A. Hersh, Sexual Morality: The Cultures and Emo-
tions of Conservatives and Liberals, 31 Journal of Applied Social Psychology 191 (2001) (same).
“They Saw a Protest” 10
other (realistically), but not itself (naïvely), of fitting its empirical beliefs about how
the world works to its moral vision of how it should.
Citizens defeated in political
conflicts of this sort will thus face the same form of humiliation they’d suffer had
their worldviews been explicitly denigrated by culturally partisan laws. To avoid
this experience, groups predictably mobilize and energize their members by advo-
cating positions that expressively affirm their own partisan valuesthereby pro-
voking reciprocal anxiety and resistance by their adversaries, who can be expected
in turn to resort to status-protective symbolic political action.
C. Cognitive Illiberalism and the Speech-conduct Distinction
There is an obvious tension between the phenomenon of culturally moti-
vated cognition and the teleological conception of the speech-conduct distinction.
Delimiting the scope of the First Amendment requires legal decisionmakers to
determine whether a regulation (in general, and as applied in particular instances)
is justified by a governmental purpose independent of aversion to any idea expressed
by regulated acts. Such an assessment involves factual judgments akin to the em-
pirical assessments that lawmakers and citizens make in considering the utilitarian
efficacy of policies and laws: Is there a basis for believing the regulated behavior is
causing the asserted harm? Are the magnitude of the harm and the effect of the
regulation in abating it sufficiently large in relation to the cost of the regulation? Is
indifference to behaviors that cause like harms grounds to suspect the genuineness
of the regulators’ professed motivations? In making these sorts of determinations,
legal decisionmakers are thus likely to experience the same type of identity-
protective pressure that influences them to form culturally congenial perceptions
of risk and other policy-consequential facts.
The potential impact of culturally motivated cognition on facts pertinent to
the “speech”-“conduct” distinction, however, is arguably even more troubling
than its impact on perceptions of policy-consequential facts. The vulnerability of
democratic policy-making to anti-liberal impulses is familiar. It is precisely because
we anticipate that democratically accountable officials will sometimes indulge the
temptation to make law an instrument of cultural orthodoxy that we envision the
Constitution, enforced by an independent system of adjudication, as integral to
realization of liberal political principles in law. Indeed, the idea that democratically
accountable actors might sometimes unwittingly succumb to partisan temptation is
itself contemplated by the practice of judicial “strict scrutiny,” which probes the
proffered justification of laws that incidentally abridge constitutional liberties to
“flush out” unconscious illicit intentions as well as deliberately concealed ones.
However, this critical checking function would be subverted if factfinding and
See generally Robert J. Robinson, Dachter Keltner, Andrew Ward & Lee Ross, Actual Versus As-
sumed Differences in Construal: "Naive Realism" in Intergroup Perception and Conflict, 68 J. Personality & Soc.
Psych. 404 (1995).
See Dan M. Kahan,
The Cognitively Illiberal State, 60 Stan. L. Rev. 115 (2007). See generally Joseph R.
Gusfield, Symbolic Crusade: Status Politics and the American Temperance Movement (2d ed. 1986).
See, e.g., Kagan, supra note 7, at 431 n.55.
“They Saw a Protest” 11
other elements of constitutional review were themselves subject to unwitting cor-
ruption by cognitive illiberalism.
Is this a psychologically realistic concern? Cultural cognition has already been
shown to exert an impact on perceptions of legally consequential facts very similar
to the one it exerts on perceptions of risk. Issues such as “consent” in acquain-
tance rape cases,
the risks posed by fleeing suspects against whom the police use
deadly force,
and the feasibility of nonlethal alternatives when battered women
and other controversial offenders resort to homicidal violence in self-defense
ripe with social meanings. Studies show that citizens of diverse cultural outlooks
divide along predictable lines when assessing such facts. This evidence furnishes
reason to worry that fact-finding essential to constitutional law will be similarly
pervaded by culturally motivated cognition.
But conjecture and story-telling, as suggestive of hypotheses as they might
be, are not a substitute for proof.
The most reliable way to examine the potential
impact of culturally motivated cognition on the “speech-“conduct” distinction is
to conduct an empirical study of the possibility.
A. Overview and Hypotheses
We conducted a study to test the hypothesis that cultural motivated cognition
will distort perception of the line between “speech” and “conduct.” The study
focused on the lawfulness of police action to halt a political demonstration for
allegedly obstructing, threatening, or intimidating members of the public. In broad
outlines, this is a recurring scenario across diverse settings, from anti-war rallies to
pro- and anti-civil rights marches to the picketing of commercial establishments,
court houses, foreign embassies, and abortion clinics.
See Dan M. Kahan,
Culture, Cognition, and Consent: Who Perceives What, and Why, in 'Acquaintance Rape'
Cases, 158 U. Pa. L. Rev. 729 (2010).
See Dan M. Kahan, David A. Hoffman & Donald Braman,
Whose Eyes Are You Going to Believe?
Scott v. Harris and the Perils of Cognitive Illiberalism, 122 Harv. L. Rev. 837 (2009).
See Dan M. Kahan & Donald Braman,
The Self-defensive Cognition of Self-defense, 45 Am. Crim. Law.
Rev. 1 (2008).
Sood and Darley, supra note 3, report that individuals are likely to impute “harm” to behavior they
find offensive when told that only “harmful” behavior can be criminalized, a finding, they recognize,
with implications for constitutional law. Our study compliments theirs both by connecting motivated
reasoning to the specific facts relevant to distinguishing permissible regulations of “conduct” from
impermissible regualtions of “speech,” and by examining how motivated cognition interacts with
diverse systems of values, the distinctive focus of cultural cognition.
See generally Jeffrey J. Rachlinski, Comment: Is Evolutionary Analysis of Law Science or Storytelling?, 41
Jurimetrics 365 (2001); Dan M. Kahan, The EconomicsConventional, Behavioral, and Politicalof “Subse-
quent Remedial Measures” Evidence, 110 Col.umbia L. Rev. 1616 (2010).
“They Saw a Protest” 12
First Amendment jurisprudence here reflects the teleological conception of
the speech-conduct distinction. The state’s obligation to permit “expression of
unpopular views” rules out the enforcement of any governmental interest that is
related to the “communicative content” of protest activity, such as protecting tar-
gets of criticism from the “indignity” of public opprobrium,
shielding parties
from the “inconvenience [or] annoyance” of having to avoid disagreeable ideas,
or forestalling unrest caused by onlookers’ own aversion to the message protectors
are conveying.
“The right to attempt to persuade others to change their views
. . . may not be curtailed simply because the speaker’s message may be offensive to
his audience.”
Indeed, because “a function of free speech under our system of
government is to invite dispute,” it is to be expected that it will sometimes “in-
duce[] a condition of unrest, create[] dissatisfaction with conditions as they are, or
even stir[] people to anger.”
Nevertheless, the police needn’t stand idly by
“when. . . the speaker passes the bounds of argument or persuasion and under-
takes incitement to riot.”
Nor does the First Amendment prevent the police
from intervening to stop demonstrators from engaging in assaultive behavior such
as “jostling, grabbing, pushing, and shoving” or from intimidating others through
‘in your face’ yelling.”
Discharging the “responsibility to keep the[] streets
open and available for movement” and to assure passersby “entrance to a public
or private building” also justifies police action to terminate a political demonstra-
Yet in all cases, it is necessary to scrutinize the facts to assure that the as-
sertion of the “interest of the community in maintaining peace and order on its
is not used to disguise censorial motives on the part of either the au-
thorities or the public.
Boos v. Barry, 485 U.S. 312, 322 (1988).
Terminiello v. Chicago, 337 U.S. 1, 4 (1949).
See, e.g., Cox v. Louisiana, 379 U.S. 536, 550 (1965) (“the fear of violence . . . based upon the reac-
tion of” angry onlookers not sufficient to justify breaking up a civil rights demonstration); Texas v.
Johnson, 491 U.S. 397, 408-09 (1989) (“The State’s position, therefore, amounts to a claim that an
audience that takes serious offense at particular expression is necessarily likely to disturb the peace
and that the expression may be prohibited on this basis. Our precedents do not countenance such a
Hill v. Colorado, 530 U.S. 703, 716 (2000).
Terminello, 337 U.S. at 4.
Feiner v. New York, 340 U.S. 315, 321 (1951).
Schenck v. Pro-Choice Network, 519 U.S. 357, 363 (1997).
Cox, 379 U.S. at 555.
Feiner, 340 U.S. at 320.
See, e.g., Cox, 379 U.S. at 546-47 (“The State argues . . . that while the demonstrators started out to
be orderly, . . . [their behavior thereafter] converted the peaceful assembly into a riotous one. The
record, however, does not support this assertion.”); Edwards v. South Carolina, 372 U.S. 229, 234
(1963) (“The state courts have held that the petitioners’ conduct constituted breach of the peace
under state law, and we may accept their decision as binding upon us to that extent. But it neverthe-
“They Saw a Protest” 13
In our study, subjects were instructed to imagine they were jurors in a case
that turned on whether a group of protestors had crossed the speech-conduct line
so conceived. The subjects indicated their findings on key facts after viewing a
videotape of a political demonstration that was (we told them) halted by the po-
lice. The use of a video was designed to enhance the realism of the design. Cases
challenging the use of police authority to halt allegedly violent, intimidating, or
disorderly demonstrations often feature videotapes of the demonstrators’ beha-
When such cases are reviewed by appellate courts, moreover, judges some-
times disagree with each other about whether the video depicts protected speech
or instead regulable conduct.
To sharpen exploration of how values affect such perceptions, our study in-
volved an experimental manipulation. Half of the subjects were advised that the
filmed demonstration occurred outside an abortion clinic, and was aimed at pro-
testing legalized abortion (“abortion clinic condition”); the other half were told the
demonstration occurred outside of a college career-placement facility during inter-
views by the military, and was aimed at protesting the armed forces’ then-existing
ban on service by openly gay and lesbian soldiers (“recruitment center condi-
In both conditions, subjects were advised that the protestors were suing
the police for ordering the protestors to disperse on the basis of an ordinance
prohibiting “obstructing,” “intimidating,” and “threatening” persons seeking to
use the facilities in question. The design permitted us to examine, first, whether
subjects with opposing cultural worldviews would form different fact perceptions
when they were assigned to the same experimental condition (that is, when they
had the same beliefs about the cause of the demonstrators); and second, whether
subjects assigned to one condition would form fact perceptions at odds with those
of subjects who shared their worldview but who were assigned to the other condi-
tion (that is, who had a different belief about the cause of the protestors).
less remains our duty in a case such as this to make an independent examination of the whole
See, e.g., Madsen v. Women’s Health Center, 512 U.S. 753 (1994) (abortion-clinic protest); Cox v. Loui-
siana, 379 U.S. 536 (1965) (civil rights protest); United States v. Soderna, 82 F.3d 1370, 1373 (7th Cir.
1996) (abortion-clinic protest); Jones v. Parmley, 465 F.3d 46 (2d Cir. 2006) (tax protest).
See, e.g., Cannon v. City of Denver, 998 F.2d 867 (10th Cir. 1993) (overturning ruling of summary
judgment against abortion-clinic protestors suing police for breach-of-peace arrest: “We find in the
instant case no assault or threatening of bodily harm, no truculent bearing, no intentional discourte-
sy, no personal abuse,” but rather “only an effort to persuade a willing listener”) with id. at 880 (An-
derson, J., separately concurring) (“Frankly, in my view if the plaintiffs' evidence at the end of a trial
remained as it now stands, the trial judge would be entitled to grant a defense motion . . . for judg-
ment as a matter of law” and thus “the record before us now could support the grant of summary
judgment to the defendant”).
The study was conducted in November 2010, before congressional repeal of the military’s “don’t
ask, don’t tell” policy.
“They Saw a Protest” 14
To measure the subjects’ worldviews, we employed scales used in previous
studies of cultural cognition.
These scales characterize worldviews along two
orthogonal dimensions. The first, Hierarchy-Egalitarianism, measures the subjects’
orientations toward social orderings that either feature or eschew stratified roles
and forms of authority. The second, Individualism-Communitarianism, measures
their orientations toward orderings that emphasize individual autonomy and self-
sufficiency, on the one hand, and those that emphasize collective responsibilities
and prerogatives, on the other. Combining the two scales generates four sets of
worldviews“hierarchy individualism,” “hierarchy communitarianism,” “egalita-
rian individualism” and “egalitarian communitarianism,” to which individuals’ af-
finities can be measured with continuous worldview scores.
Based on the nature of these cultural predispositions and on previous re-
search, we formed a set of discrete hypotheses. We enumerate them and assign
each a descriptive label to facilitate exposition.
1. EI inversion. Relatively egalitarian individualist subjects, we surmised,
would form anti-demonstrator fact perceptions in the abortion condition but pro-
demonstrator perceptions in the recruitment center condition. Egalitarian individual-
ists are morally opposed both to social stratification, such as that associated with
traditional gender roles, and to institutional rankings, such as those that pervade
the military. Accordingly, we anticipated that egalitarian individualists would likely
be hostile to protestors in the abortion-clinic condition and sympathetic to those
in the recruitment-center condition. We also expected that egalitarian individualist
subjects would also feel their worldviews were being affirmed and threatened, re-
spectively, by the abortion-clinic and military-recruitment-center “free access” or-
dinances. We therefore predicted these subjects would be inclined to perceive the
protestors had engaged in prohibited “conduct” in the abortion-clinic condition
but protected “speech” in the recruitment-center condition.
2. HC inversion. We predicted that relatively hierarchical and communitarian
subjects, by contrast, would form pro-demonstrator fact perceptions in the abortion
condition but anti-demonstrator perceptions in the recruitment center condition.
Hierarchical communitarians are strongly supportive of traditional gender norms,
and as a result attach a negative social meaning to abortion rights, which to them
denigrate the status properly afforded women for successful mastery of female
domestic roles centering on maternity. We anticipated that they would therefore
find the ordinance securing free access to abortion clinics particularly obnoxious.
In contrast, they attach positive meanings to the military as an institution that is
characterized by stratified internal orderings that subordinate the individual to the
collective, and as a setting in which men, in particular, can occupy roles that dis-
play the virtue of patriotism. These resonances, we predicted, would create identi-
ty-protective pressure on hierarchical communitarian subjects to perceive the anti-
abortion demonstrators engaged in protected “speech” and the anti-military de-
monstrators engaged in obstruction, intimidation and like prohibited “conduct.”
See Dan M. Kahan,
Cultural Cognition as a Conception of the Cultural Theory of Risk, in Handbook of
Risk Theory (ed. S. Roeser, forthcoming 2011) (Springer Publishing).
“They Saw a Protest” 15
3. HI bias. We anticipated that subjects holding relatively hierarchical and in-
dividualistic values would form strong anti-demonstrator fact perceptions in the
recruitment-center condition, but more muted anti-demonstrator perceptions in
the abortion-clinic condition. Virtues such as courage, honor, and martial prowess
figure conspicuously in this way of life and are status-conferring for men in partic-
ular. We expected, then, that hierarchical individualists would be morally hostile to
the aims of the protestors in the military-recruitment condition, and hence form
fact perceptions consistent with the finding that they engaged in prohibited “con-
duct” rather than protected “speech.” We anticipated that hierarchical individual-
ists’ would be unlikely to take offense at the message of the anti-abortion protes-
tors. Nevertheless, abortion rights do not bear a meaning nearly as threatening to
hierarchical individualists as they do to hierarchical communitarians; in addition,
hierarchical individualists tend to place a high value on social order generally. We
anticipated, then, that hierarchical individualists would form less strong pro-
demonstrator perceptions in the abortion-clinic condition than would hierarchical
4. EC bias. We hypothesized that relatively egalitarian communitarian sub-
jects would form strong pro-demonstrator fact perceptions in the recruitment-
center condition and modestly anti-demonstrator fact perceptions in the abortion-
clinic condition. Egalitarian communitarians see the imposition of legal disabilities
on gays and lesbians as symbols of institutionalized patriarchy. They strongly sup-
port gay marriage and gay parenting, the social meanings of which enable alterna-
tive, nonpatriarchal forms of community and shared commitment.
We antic-
ipated that similar sensibilities would make them supportive of lifting removal of
restrictions on military service by openly gay and lesbian citizens, and hence trigger
culturally motivated cognition supportive of the recruitment-center protestors. We
also expected that egalitarian communitarians would be offended by the anti-
gender equality resonances of the abortion clinic protestors. Nevertheless, abor-
tion rights also bear individualistic meanings that egalitarian communitarians resist.
Accordingly, we anticipated that egalitarian communications in the abortion-clinic
condition would feel less impelled than egalitarian individualists in that condition
to perceive the demonstrators as engaged in prohibited conduct rather than pro-
tected speech.
5. EI/HC polarization. The final two hypotheses relate to the expected inten-
sity and character of the disagreement between subjects of opposing cultural iden-
tities. We hypothesized that in both conditions there would be strong, mirror im-
age forms of polarization between relatively egalitarian and individualistic subjects,
on the one hand, and relatively hierarchical communitarian ones, on the other.
This prediction was simply a logical implication of the EI and HC inversion hypo-
6. EC/HI semi-polarization. Consistent with the HI and EC bias hypotheses,
we predicted that disagreement between egalitarian and communitarian subjects,
See Cultural Cognition Project, First Report on Gay & Lesbian Parenting, available at
“They Saw a Protest” 16
on the one hand, and hierarchical and individualistic ones, on the other, would be
less symmetric. We expected the two to be strongly polarized in the recruitment-
center condition, in which the cultural meaning of the protestors’ cause would
exert diametrically opposing forces on their respective perceptions. However, be-
cause abortion rights bear more equivocal meanings within the worldviews of both
of these groups, we anticipated that their disagreement in the abortion-clinic con-
dition would likely be more moderate.
Figure 1. Summary of hypotheses. Based on their predispositions, subjects with oppos-
ing cultural values were expected to disagree with each other within each experimental
condition, while those with the same values are expected to disagree with one another be-
tween experimental conditions. The EI/HC polarization hypothesis predicted that egalitarian
individualists and hierarchical individualists would form diametrically opposed perceptions
in both conditions. The EC/HI semi-polarization hypothesis predicted that egalitarian commu-
nitarians and hierarchical individualists would polarize most intensely in the military-
recruitment-center condition.
B. Design and Methods
1. Sample
The subjects for the study consisted of 202 American adults. They were se-
lected randomly from a stratified national sample by Polimetrix, Inc.,
and parti-
cipated in the study through Polimetrix’s on-line testing facilities. Forty-six percent
of the sample was female. Seventy-two percent were white, and nine percent Afri-
can-American. The median level of education was between “some” and “two
years” of college. The median annual income was between $40,000 and $49,999.
The average age was 46.
Polimetrix ( is a public opinion research firm that conducts on-line
surveys and experiments on behalf of academic and governmental researchers and commercial cus-
tomers (including political campaigns). It maintains a panel of over 1 million Americans that it uses
to construct representative study samples. For more information, see
“They Saw a Protest” 17
2. Cultural Worldviews
Subjects’ cultural worldviews were measured (in advance of the study) with
the “Hierarchy-egalitarianism” (“Hierarchy”) and Individualism-communitarianism
(“Individualism”) scales used in previous studies of cultural cognition.
scales consisted of twelve statements expressing attitudes characteristic of one or
the other worldview dimension (e.g., “Society as a whole has become too soft and
feminine”; “The government interferes far too much in our everyday lives”), and
subjects indicated agreement or disagreement on a six-point scale. Each six-item
scale was highly reliable (Hierarchy: Cronbach α = 0.87; Individualism: Cronbach’s
α = 0.81),
and the twelve items loaded appropriately on two separate factors,
which were used as continuous predictors for multivariate testing of the study hy-
In addition, to enable illustrative analyses, we designated each subject
either a “hierarchical individualist,” a “hierarchical communitarian,” an “egalitarian
individualist,” or an “egalitarian communitarian” based on his or her scores in rela-
tion to the sample medians on each scale.
3. Stimulus
Subjects were randomly assigned to either the “abortion clinic” condition or
the “recruitment center” condition. They were then assigned to read a vignette and
view an accompanying video.
a. Vignette. The vignette described the background of a lawsuit by political
protestors against individual police officers and the police department. Depending
See, e.g., Dan M. Kahan, Hank Jenkins-Smith & Donald Braman,
Cultural Cognition of Scientific Con-
sensus, 14 J. Risk Res. 147 (2011). For a full discussion of the complete and short-form versions of
the scales and of their psychometric properties, see Dan M. Kahan,
Cultural Cognition as a Conception of
the Cultural Theory of Risk, in Handbook of Risk Theory (ed. S. Roeser, forthcoming 2011) (Springer
Cronbach’s alpha (α) is a statistic for measuring the internal validity of attitudinal scales. By com-
puting the degree of inter-correlation that exists among various items within a scale, it can be used to
assess whether the items can properly be treated as common indicators of a latent attitude or trait
i.e., one that cannot be directly observed and measured. See generally Jose M. Cortina, What Is Coefficient
Alpha: An Examination of Theory and Applications, 78 J. Applied Psychol. 98 (1993). Composite scales of
this sort are desirable not only because they facilitate measurement of unobservable dispositions but
also because the measurements they enable are necessarily more precise than ones based on any of
the individual indicators alone, each of which can be seen as an imperfect or “noisy” approximation
of the phenomenon being studied. See generally J. Philippe Rushton, Charles J. Brainerd & Michael
Pressley, Behavioral development and construct validity: The Principle of Aggregation, 94 Psychol. Bull. 18
(1983). Generally, α .70 suggests scale validityi.e., that the measures when aggregated furnish a
reliable measure of the latent trait or attitude. See Cortina, supra.
Treating Hierarchy and Individualism as continuous predictors maximizes statistical power and avoids
the bias that can be introduced by splitting them at the mean or other selected points in order to
transform them into discrete, categorical measures. See James Jaccard & Robert Turrisi, Interaction
Effects in Multiple Regression 86 (2nd ed. 2003).
The study instrument, including the vignette and response instruments is reproduced in Appendix
A. The videos can be viewed on-line at and
“They Saw a Protest” 18
on the condition, the protestors were described either as “members of a group
that opposes permitting doctors and nurses to perform abortions at the request of
pregnant women” or as “members of a group that opposes the ban on allowing
openly gay and lesbian citizens to join the military.” The protestors’ complaint, the
vignette stated, alleged that the police had “violated their rights by ordering them
to end their protest at” either “an abortion clinic” or “a college campus recruit-
ment center the day the Army was scheduled to interview students who were con-
sidering enlisting.”
Subjects were told that the defendants claimed halting the protest was justi-
fied by a law entitled the “Freedom to Exercise Reproductive Rights Law,” in one
condition, or the “Freedom to Serve with Honor Law,” in the other. The law,
enacted after a previous judicial ruling found that police lacked “clear guidelines”
for halting such protests, made it illegal for “any person to intentionally
(1) interfere with
, (2) obstruct, (3) intimidate, or (4) threaten any person who is
seeking to enter, exit, or remain lawfully on premises of” either “any hospital or
medical clinic that is licensed to perform abortions” or “any facility in which the
U.S. military is engaged in recruitment activity.” This text was patterned on the
federal Freedom of Access to Clinics Entrances Act,
enacted in response to
demonstration activity perceived to be intended to impede operation of abortion
facilities, and the 2008 “Freedom to Serve Act” bill,
which would have created a
similar provision relating to military recruitment facilities.
Each version also au-
thorized officers to order protestors to cease and leave the vicinity upon “ob-
serv[ing] or [being] furnished with reliable evidence” that the law was being vi-
olated. The protestors, according to the vignette, alleged that they had been “only
expressing their views, in a manner that did not violate the law.”
Subjects were advised that both parties agreed that a “video of the protest”
furnished an “accurate impression of the nature of the protestors conduct” and
18 U.S.C. § 248(a)(1) (“FACE”) (“Whoever by force or threat of force or by physical obstruction,
intentionally injures, intimidates or interferes with or attempts to injure, intimidate or interfere with
any person because that person is or has been, or in order to intimidate such person or any other
person or any class of persons from, obtaining or providing reproductive health services” is subject
to criminal penalties).
H.R. 6023, § 3, 110th Congress, May 12, 2008 (“Whoever by force or threat of force or by physi-
cal obstruction, injures, intimidates or interferes with or attempts to injure, intimidate or interfere
with any person because that person is or has been providing Federal or State military recruiting
In Hill v. Colorado, 530 U.S. 703 (2000), all nine Justices, including three dissenters who would
have invalidated the statute on other grounds, endorsed the constitutionality of language similar to
that in our vignettes. See id. at 754-55 (Scalia, J., joined by Thomas, J.) (portion of law that “knowing-
ly obstructs, detains, hinders, impedes, or blocks another person’s entry to or exit from a health care
facility” is “narrowly tailored to serve” the state’s “asserted” interest in securing access to such a
facility and would not have been reinforced with additional provisions had the state not also interest-
ed in stifling abortion-clinic protestors in particular); id. at 777 (presenting the same position). Lower
courts have rejected constitutional challenges to FACE. See, e.g., United States v. Soderna, 82 F.3d
1370, 1375 (7th Cir. 1996) (Posner, J.) (“To persuade and to blockade are importantly different forms
of action.”).
“They Saw a Protest” 19
thus represented “the key evidence” in the case. However, the parties were de-
scribed as
disagree[ing] about whose position the video most supports: the posi-
tion of the police officers, who assert that the protestors were “intimi-
dating, interfering, obstructing or threatening” people trying to enter the
abortion clinic campus recruitment center; or the position of the protes-
tors, who say they were merely expressing their views in a lawful man-
“Deciding who is right is the task for you as a member of the jury,” the vignette
stated. Subjects were then instructed to view the video.
b. Video. Approximately three and one-half minutes in length, the video de-
picts an actual political demonstration that occurred in Cambridge, Massachusetts,
in March, 2009. The protestors included approximately half-a-dozen members of
the Westboro Church, a Kansas-based group whose members conduct demonstra-
tions condemning homosexuality.
Also present was a substantially larger number
of counter-demonstrators (approximately 200), although the video was designed to
create the impression that they and the church members formed a single mass of
protestors. The video consisted of five distinct scenes showing both the Westboro
Church members and the counter-demonstrators congregated near the entrance of
a building. The video also contains numerous shots of helmeted police officers
who were present to direct traffic and control the crowd in the vicinity of the pro-
test. In certain scenes, pedestrians (all college-aged males and females) are shown
either veering away from the protestors gathered near the entrance of the building
or walking in the opposite direction of the entrance while looking over their
shoulders at the crowd (Figure 2).
See Snyder v. Phelps, 113 S. Ct. 1207 (2011). The Supreme Court granted certiorari in Phelps approx-
imately a year after the Cambridge demonstration that is featured in the study videotape. At issue in
Phelps was an award of damages against the Westboro Church members for intentional infliction of
emotional distress. The basis of the award was a protest that the Church conducted at the funeral of
a soldier, whose death, the Church asserted, was an act of retaliation by God for the United States’s
tolerance of homosexuality. Applying the teleological conception of speech, the Court held that the
award of damages to the soldier’s father violated the First Amendment because “any distress occa-
sioned by Westboro’s picketing turned on the content and viewpoint of the message conveyed, ra-
ther than any interference with the funeral itself.” Id. at 1219. The Court emphasized the “narrow”
basis of this disposition, id. at 1220; nothing in its ruling, the Court indicated, would affect the au-
thority of the state to sanction or prohibit protest behavior that was “unruly” or “violen[t],” id. at
1219-20, or that otherwise “dirsupt[ed] th[e] funeral,” id. at 1220the sorts of “noncommunicative”
harms that can be averted or redressed consistent with the First Amendment.
“They Saw a Protest” 20
Figure 2. Video. Subjects in each condition viewed a video of a political demonstration.
Wording of signs was blurred to prevent identification of the actual subject matter of the
protest. Subjects were instructed that blurring had been ordered by the court to prevent
jurors from being influenced by the messages they contained.
A screen of explanatory text appeared before each scene (Figure 3). De-
scribed as “based on witness statements the parties agree is accurate,” the text re-
lated, in a deliberately bland manner meant to avoid expressing a position on any
disputed issue, what the next scene would show (e.g., “Outside [the reproductive
health clinic or campus recruitment center] 15 minutes before it was scheduled to
open”). The text indicated that the pedestriansdescribed as either clinic “pa-
tients” and “staff” or “students” scheduled for interviewsdid not enter the facil-
ity but did not state a reason (“Scene at the entrance of [clinic/campus recruitment
center]. [Patient/student] approaches but does not enter”). In two scenes, a mid-
dle-aged man conversing with a police officer is identified as the “director” of ei-
ther the “clinic” or “recruitment center” and is described as urging the police to
halt the demonstration.
Modifications of the film were also made to prevent subjects from identifying
the actual positions of either the Westboro Church members or the counterde-
monstrators. These included the blurring of messages on their respective signs;
subjects were instructed that the court had ordered the blurring to “assure that
th[e] messages did not affect the jury one way or the other,” because “the U.S.
Constitution prohibits the police from breaking up a protest based on the messages
the protestors are trying to communicate.” Generic crowd noise, consisting pri-
marily of a cacophony of shouts and chants, was added as a sound track.
“They Saw a Protest” 21
Figure 3. Video text screens. Each of the five scenes in the video was introduced by a text
screen. The text screens contained minor variations to fit the experimental condition but
were otherwise identical in both conditions.
A pretest conducted on a group of approximately 100 judges and lawyers
confirmed that the tape could be plausibly identified as either an abortion-clinic or
a recruitment-center protest. None of the participants in the pretest recognized the
protestors. Debriefing feedback for the study suggested that only one subject iden-
tified the protestors as members of the Westboro Church. That subject’s res-
ponses were therefore excluded from analysis.
The intent of both the filming and editing was to create grounds for oppos-
ing conclusions about the key facts. At no point is there physical contact between
the protestors or counter-demonstrators and the pedestrians identified as not en-
tering the facility. Nevertheless, the proximity of the protestors to the pedestrians
and their proximity to the entrance would have furnished a basis for inferring that
the pedestrians were either obstructed or intimidated. So would the passionate
behavior of the demonstrators, including in particular one female protestor who is
shown at various points yelling and gesturing in the direction (one would assume)
of those intent on entering the facility. Yet it could also have been inferred that
pedestrians avoided entering either because they were persuaded by the protestors’
message, were averse to being obliged to listen to the protestors, or were anxious
not to be publicly condemned for their behavior. Members of the lawyer-judge
pretest panel were close to evenly divided on these matters and thus on whether
the police had cause for ordering a cessation of the protest.
In reality, the police did not halt the protest, which terminated without incident after a period of
approximately 45 minutes.
Recruitment Center Condition
Abortion Clinic Condition
“They Saw a Protest” 22
4. Response Measures
After watching the video, subjects were asked to indicate on a six-point scale
their level of disagreement or agreement with twenty-one response items. The first
seventeen items related to various facts relevant to application of the standard set
forth in “the Freedom to Exercise Reproductive Rights Law” or the “Freedom to
Serve with Honor Law.” Some of the items indicated that the police acted on
grounds that would justify treating the protest activity as prohibited “conduct”
under prevailing First Amendment doctrine (e.g., “The protestors obstructed indi-
viduals seeking to enter, exit, or remain lawfully on the premises of the [abortion
clinic/campus recruitment center”; “There was a risk that the protestors might
resort to violence if anyone tried to enter.”).
Others, in contrast, suggested the
protestors were engaged in lawful “speech” and that the motivation for ordering a
halt to the demonstration was “speech related” and hence impermissible (e.g.,
“The protestors intended only to persuade people not to go into the [abortion
clinic campus recruitment center], not to physically interfere with, intimidate, ob-
struct, or threaten anybody”; “It is more likely the director asked the police to
break up the protest because the director and others found dealing with the pro-
testors annoying than because the protestors were interfering with, intimidating,
obstructing, or threatening anyone”).
Some items attributed to the protestors
behavior that clearly did not occurthat they “shoved” prospective facility users
and “spit” at them, for examplebut most of the facts were matters of interpreta-
tion and inference.
The last four items related to the proper disposition of the case. These in-
cluded the appropriateness of an award of damages and entry of an injunction
against future police “interfer[ence] with protests under conditions like the ones
shown in the video.”
The fact and case-disposition items formed a highly reliable composite scale.
Designated Pro_Protest (Cronbach’s α = 0.95), the scale furnished a continuous
measure (standardized by z-score transformation) of each subject’s relative inclina-
tion to form pro-demonstrator fact perceptions and case-disposition judgments.
The necessity of finding facts such as these, which parallel ones that the Supreme Court has indi-
cated supply permissible grounds for restricting protest activity, see supra pp. 11-12, has been empha-
sized by the lower courts that have rejected constitutional challenges to the Freedom of Access to
Clinic Entrances Act. See, e.g., United States v. Scott , 958 F. Supp. 761, 764 (D. Conn. 1997). The court
in Scott found that one defendant protestor had violated FACE by “physically obstructing and threat-
ening persons seeking to enter and exit . . . in order to intimidate and interfere,” id. at 768, but that a
second had not because “her intent in sidewalk counseling and leafleting [was only] to persuade
women to consider alternatives to abortion,” id. at 770-71.
Cf. Terminello, 340 U.S. at 4.
Items expressing a pro-defense fact perception or outcome-judgment were reverse coded.
“They Saw a Protest” 23
5. Analytic Strategy
Multivariate regression furnishes an appropriate and straightforward proce-
dure for testing the study hypotheses.
Because Pro_Protest supplies a more pre-
cise measure of the latent disposition to form pro-plaintiff reactions than do the
individual items, the scale furnishes the most reliable outcome variable for testing
the study hypotheses.
The predictors in our study include the experimental ma-
nipulation, which we denote by the variable Recruitment (abortion-clinic = 0; re-
cruitment-center = 1); and subjects’ worldviews, which are measured by their
scores on the Hierarchy and Individualism scales. To enable testing of hypotheses
relating to the varying impact of subjects cultural worldviews in the two experi-
mental conditions, we constructed product-interaction terms: Hierarchy_x_Re-
cruitment and Individualism_x_Recruitment, which measure any difference that
an increasing disposition toward hierarchy or individualism, respectively, has on
subjects’ fact and case case-disposition responses in the recruitment-center (Re-
cruitment = 1) as opposed to the abortion-clinic (Recruitment = 0) condition.
C. Results
The study results appear in the regression analyses reported in (Table 1). Pre-
dictors and cross-product interaction terms are entered in steps to promote inter-
pretation of the contribution that the various predictors make to variance in reac-
tions to the video, as measured in Pro_Protest.
Model 3 incorporates all the pre-
dictor and cross-product interaction terms that bear on the study hypotheses.
Putting aside subjects’ cultural worldviews, the impact of being assigned to
one experimental condition or another is negligible. The coefficient for Recruitment
indicates the impact of being assigned to the recruitment-center as opposed the
abortion-clinic condition. In Model 1, in which the experimental assignment is
treated as the only predictor, the value of the coefficient for Recruitment is close
to zero and is statistically nonsignificant.
Charles M. Judd, Everyday Data Analysis in Social Psychology: Comparisons of Linear Models, in Hand-
book of research methods in social and personality psychology 370 (H. T. Reis and C. M. Judd ed.,
2000) (outlining use of multivariate regression for analysis of experimental results and explaining
advantages over ANOVA).
See note 73 supra.
See generally Jacob Cohen, Patricia Cohen, Stephen G. West & Leona S. Aiken, Applied Multiple
Regression/Correlation Analysis for the Behavioral Sciences 375-83 (3rd ed. 2003) (outlining and
explaining the use of cross-product interaction terms in multivariate regression to model and test the
hypothesis that the effect of a continuous predictor will vary across the levels of a categorical one).
The underlying regression equation in Model 3 is
Y = b1 x Recruitment + b2 x Hierarchy + b3 x Individualism + b4 x Hierarchy x Recruitment + b5
x Individualism x Recruitment + constant
where Y is the score on the score on Pro_Protest, and b1-b5 the coefficients for the specified predic-
tors and cross-product interaction variables.
“They Saw a Protest” 24
Model 1
Model 2
Model 3
Hierarchy x Recruitment
Individualism x Recruitment
Table 1. Multivariate regression analysis. N = 196. Dependent variable is Pro_Protest,
the composite response-item scale transformed to a z-score with a mean of 0. Predictor
effects are measured with unstandardized linear (OLS) regression coefficients. Standard
errors are in parentheses. Bold type indicates that the predictor coefficient or change in F
statistic is significant at p < 0.05.
Aggregate impressions not only were comparable between the two condi-
tions but were also closely divided within each. This conclusion is reflected clearly
in the raw data (Figure 4). In the abortion-clinic condition, 49% of the subjects
indicated that they agreed (either “slightly,” “moderately,” or “strongly”) that the
police should be found liable for ordering the protestors to cease the demonstra-
tion. In the recruitment-center condition, 45% of the subjects agreed the police
should be found liable (Figure 4). Comparable proportions supported the proposi-
tion that the police should be enjoined from halting protests “under conditions
like the ones shown in the video” (recruitment center, 46%; abortion clinic, 45%).
The proportion who agreed that the protestors should be awarded damages was
smaller25% in each condition.
Figure 4. Main effects. Bars indicate the percentage of subjects who agreed (either
“slightly,” “moderately,” or “strongly”) with items proposing the indicated dispositions.
Considered apart from the experimental manipulation, cultural worldviews
likewise appear to have no meaningful effect on reactions to the video. The coeffi-
cients for Hierarchy and Individualism in Model 2 indicate the impact of subjects’
recruitment ctr
recruitment ctr
recruitment ctr
abortion clinic
abortion clinic
abortion clinic
Police liable Grant Injunction vs.
Damages vs. police
Pct. Agree
“They Saw a Protest” 25
scores on the indicated worldview variables averaged across the two conditions.
Again, both coefficients are close to zero and are statistically nonsignificant. There
is no evidence, then, that being inclined either toward hierarchy or egalitarianism,
toward individualism or communitarianism, or toward any combination of the two
disposes individuals toward pro- or anti-demonstrator reactions irrespective of
what subjects believe about the political cause of the demonstrators.
When cultural worldviews and experimental conditions are considered togeth-
er, however, it becomes clear that who saw what did depend critically on the rela-
tionship between the demonstrators’ causes and the subjects’ own values. The na-
ture of these influences, moreover, was consistent with study hypotheses.
The condition-specific effect of each worldview is reflected in Model 3.
Their impact in the abortion-clinic condition are indicated by the coefficient for
Hierarchy, which is positive, and by the coefficient for Individualism, which is
negative (both are significant).
These results indicate that subjects in the abor-
tion-clinic condition formed progressively more pro-demonstrator fact perceptions
and case-disposition preferences as their values became either more hierarchical or
more communitarian; by the same token, they formed progressively more anti-
demonstrator perceptions and preferences as their values became either more egali-
tarian or more individualistic. These relationships are reversed in the recruitment-
center condition: the positive coefficients for Hierarchy_x_Recruitment and the
negative ones for Individualism_x_Recruitment indicate that in that condition pro-
demonstrator reactions dissipated as subjects become either more hierarchical or
communitarian, but grew as subjects become either more egalitarian or individua-
It follows that subjects who are simultaneously more egalitarian and indivi-
dualistic will form relatively extreme anti-demonstrator impressions in the abor-
tion-clinic condition and pro-demonstrator impressions in the recruitment-center
condition. Subjects who are who are simultaneously more hierarchical and com-
munitarian (under any of the models) will form exactly the opposite impressions.
These results thus confirm the EI inversion and HC inversion hypotheses, and hence
the EI/HC polarization hypothesis as well.
It is more difficult to assess the remaining hypotheses by simply scrutinizing
the regression outputs. Because Hierarchy and Individualism have opposite signs
In Model 3, the coefficient for Recruitment and the constant are the effects of being assigned
either to the military-recruitment center condition or to the abortion-clinic condition, respectively,
when the cultural worldviews are equal to zero, or their mean values. Again, they are close to zero
and statistically nonsignificant. One can thus conclude that the “culturally average” subject would
react comparably in both conditions.
The coefficients for each worldview predictor indicates the impact of the specified worldview
scale when all the other predictors equal zero, see Leona S. Aiken, Stephen G. West & Raymond R.
Reno, Multiple Regression: Testing and Interpreting Interactions 123-25 (1991), which will be true
when a subject is assigned to the abortion-center condition (Recruiment = 0) and has the mean score
on the other worldview scale.
The coefficients for each cross-product interaction variable indicates the unique incremental ef-
fect associated with the indicated worldview in the recruiment-center condition. See generally id.
“They Saw a Protest” 26
from each other in both conditions (and for each outcome variable), disagreement
between subjects who are more hierarchical and individualistic, on the one hand,
and those who are more egalitarian and communitarian, on the other, will neces-
sarily be less extreme. This is consistent with the EC/HI semi-polarization hypothe-
sis, but cannot be determined to support either it or the EC bias and HI bias hypo-
theses unless the magnitudes of offsetting effects are estimated and compared in
each condition.
Figure 5. Multivariate regression estimates. Graphs display the impact of culture-
condition interactions tested in the multivariate regression analyses (Table 1). Group esti-
mates formed by setting values for Hierarchy and Individualism at ± 1 SD from the mean
in the specified directions. CIs reflect 0.95 level of confidence. Change in each group’s
score between conditions is significant at p < 0.05.
Those effects are plotted in Figure 5. Consistent with the EI and HC inversion
hypotheses, the estimated scores for Egalitarian Individualists and for Hierarchical
Communitarians scales flip.
It is also evident from the estimates that Egalitarian
Communitarians became significantly more pro-demonstrator, Hierarchical Indi-
vidualists significantly more anti-demonstrator, in the recruitment-center condition
as opposed to the abortion-center condition. Whereas the difference between Ega-
litarian Individualists and Hierarchical Communitarians is significant in both con-
ditions, the difference between Egalitarian Communitarians and Hierarchical Indi-
The statistical significance of the difference between any group’s estimated score in one condition
and its estimated score in another, and of the difference between any two group’s estimated scores
within a condition, must be determined by calculation. However, a rough visual heuristic is to con-
sider whether the 0.95 confidence interval of one estimate overlaps with the point estimate of anoth-
er (not the latter’s confidence interval, as is sometimes mistakenly stated). See generally Geoff Cum-
ming & Sue Finch, Inference by Eye: Confidence Intervals and How to Read Pictures of Data, 60 Am. Psychol.
170 (2005); Sarah Belia, Fiona Fidler, Jennifer Williams & Geoff Cumming, Researchers Misunderstand
Confidence Intervals and Standard Error Bars, 10 Psychol. Methods 389 (2005); Nathaniel Schenker &
Jane F. Gentleman, On Judging the Significance of Differences by Examining the Overlap Between Confidence
Intervals, 55 Am. Stat. 182 (2001).
For purposes of these estimates, the values for the cultural worldview predictors were both set
one standard deviation from their means in the directions necessary to form the specified worldview
combinations. See id. at 13.
“They Saw a Protest” 27
vidualists is significant only in the recruitment center condition. This result fits the
EC/HI semi-polarization hypothesis, although we had not anticipated that there
would be essentially no meaningful difference whatsoever between Egalitarian
Communitarian and Hierarchical Individualistic subjects in the abortion-center
Figure 6. Case-outcome measures, by cultural group. Lines connect points indicating
the percentage of subjects within each of specified cultural groups who agreed (either
“slightly,” “moderately,” or “strongly”) with items proposing the indicated dispositions.
The practical impact of these effects is readily illustrated by examining the
responses for individual items (Figure 6 and Figure 7). Subjects characterized (on
the bias of their mean scores on the worldview scales) as Egalitarian Individualists
and Hierarchical Communitarians reacted in strong and opposite ways to the expe-
rimental manipulation (Figure 6). In the abortion-clinic condition, 70% of the Hie-
rarchical Communitarians found that the police had violated the demonstrators’
rights. Yet in the recruitment-center condition, only 16% did. Matters were the
other way around for Egalitarian Individualists: 76% of them concluded that the
police had violated the rights of the protestors in the military-recruitment condi-
“They Saw a Protest” 28
tion, yet only 28% of them took that position in the abortion-clinic condition.
These patterns fit the EI Inversion, HC Inversion, and EI/HC polarization hypotheses.
Figure 7. Responses to select fact items, by cultural group. Lines connect points that
indicate the percentage of subjects within each of specified cultural groups who agreed
(either “slightly,” “moderately,” or “strongly”) with items proposing the indicated facts.
Egalitarian Communitarians and Hierarchical Individualists also reacted as
predicted. In the abortion-clinic condition, 52% of both Egalitarian communita-
rians and Hierarchical Individualists found that the police abridged the protestors’
right to free speech. In the recruitment-center condition, the proportion of Egali-
tarian Communitarians who found a constitutional violation jumped to 71%, while
the proportion of Hierarchical Individualists who did fell to just 17%. This pattern
fits the EC and HI bias hypotheses, as well as the EC/HI semi-polarization hypothe-
There was comparable cultural dissensus over remedies. Majorities of Egali-
tarian Individualists (69%) and Egalitarian Communitarians (59%) favored enjoin-
ing the police in the recruitment center condition, whereas only minorities of these
subjects (Egalitarian Individualists, 12%; Egalitarian Communitarians, 48%) did in
the abortion-clinic condition. Minorities of both Hierarchical Communitarians
(32%) and Hierarchical Individualists (17%) favored issuing an injunction against
the police in the recruitment-center condition; in the abortion-clinic condition, the
proportion of Hierarchical Communitarians who supported an injunction rose to
65%, and the proportion of Hierarchical Individualists to an even 50%. Consistent
with the sample-wide outcome, support among subjects of all worldviews was
lower for damages than for injunctive relief. Nevertheless, for Hierarchical Com-
munitarians and Egalitarian Individualists, in particular, the proportions support-
Pct. Agree
Protestors blocked
Screamed in face
Pedestrians just not want to listen
Police just annoyed
Egal Indivd
Hier Comm
Hier Individ
Egal Comm
abortion clinic
recruitment center
abortion clinic
recruitment center
abortion clinic
recruitment center
abortion clinic
recruitment center
abortion clinic
recruitment center
abortion clinic
recruitment center
abortion clinic
recruitment center
abortion clinic
recruitment center
“They Saw a Protest” 29
ing damages shifted within and between conditions in patterns identical to the
shifts on the other outcome measures.
As one would expect, these differences in case-disposition judgments are
mirrored in the subjects’ responses to the fact-perception items. Whereas only
39% of the Hierarchical Communitarians perceived that the protestors were block-
ing the pedestrians in the abortion-clinic condition, for example, 74% of them saw
blocking in the recruitment-center condition. Only 45% of Egalitarian Individual-
ists, in contrast, saw blocking in the recruitment-center condition, whereas in the
recruitment-center condition 76% of them did. Fully 83% of Hierarchical Indivi-
dualists saw blocking in the military recruitment-center condition, up from 62% in
the abortion-clinic condition; a 56% majority of Egalitarian Communitarians saw
blocking in that condition, yet only 35% saw such conduct in the recruitment-
center condition. Responses on other itemssuch as whether the protestors
“screamed in the face” of pedestrians and whether the protestors intended only to
persuade or instead to threatendisplayed similar patterns.
Relatively few subjects reported observing “spitting” (18%) or “shoving”
(16%) by the protestors or “physical contact” (20%) between the protestors and
the pedestrians. There was also no meaningful cultural variation with respect to
these items. This result suggests that the influence of values was confined to facts
on which there was at least modest room for interpretation. It also helps to con-
firm that the subjects were not responding in a consciously biased manner in gen-
A. Summary of Results
The theoretical aim of the study was to test the hypothesis that culturally mo-
tivated cognition would influence individuals’ perceptions of facts essential to dis-
tinguishing “speech” from “conduct” for purposes of the First Amendment. The
results strongly supported this hypothesis. Our subjects all viewed the same video.
But what they sawearnest voicing of dissent intended only to persuade, or physi-
cal intimidation calculated to interfere with the freedom of othersdepended on
the congruence of the protestors’ positions with the subjects’ own cultural values.
Motivated cognition not only polarized individuals of diverse cultural out-
looks but also generated contradictions in what subjects of a shared orientation
reported seeing. Relatively hierarchical and communitarian subjects rejected the
proposition, credited by relatively egalitarian and individualistic ones, that demon-
strators were blocking access to a facility represented to be an abortion clinic; yet
when hierarchical communitarians understood the demonstrators to be objecting
to the exclusion of openly gay and lesbian citizens from the military, they agreed
the protestors were blocking access to the same buildinga claim that egalitarian
individualists now overwhelmingly dismissed. Subjects subscribing to a hierarchic-
al individualistic outlook as well as those adhering to an egalitarian communitarian
one exhibited similar shifts in perception.
“They Saw a Protest” 30
We focused on our subjects’ cultural worldviews because of the demonstrat-
ed role of these outlooks in shaping perceptions of risk and related facts relevant
to policy and law.
Our results thus suggest the utility of cultural cognition theory
for measuring the impact and explaining the sources of motivated reasoning in
constitutional decisionmaking as well.
It is reasonable to surmise that the controversies featured in this study would divide subjects
along lines in addition to cultural outlooks as we measure them. For example, women (55%) were
more likely than men (37%) to agree that the police should be enjoined from halting future demon-
strations. The proportions of both male and female subjects who supported this outcome, however,
did not differ meaningfully across conditions. Thus, cultural variance obviously cannot be attributed
or reduced to gender variance. Political party affiliation did register sensitivity to the experimental
manipulation: in the abortion-clinic condition, the majority of Democrats (57%) opposed an injunc-
tive remedy, and a majority of Republicans (62%) favored it; in the recruitment-center condition, a
majority of Republicans (67%) opposed and a majority of Democrats (61%) favored such an out-
come. Of course, this simple partisan inversion of impressions is necessarily less nuanced than the
cross-cutting shifts observed when members of the sample were classified simultaneously along the
two cultural dimensions. Moreover, among the third of the sample that did not identify themselves as
either Democrats or Republicans, there was again no difference between the abortion-clinic and
military-recruitment conditions. We are eager to add, however, that we regard the question “what has
the biggest impactculture, gender, political ideology, race, etc.?”as ill-posed. Cultural worldviews
tend to cohere with other characteristicsincluding political affiliation, gender, race, and classin
patterns that indicate the same latent predispositions the cultural worldviews by themselves measure.
See Kahan, supra note 72. When forced to chooseas one often is, by sample sizecultural
worldviews can be expected to be more discerning indicators of these predispositions, and hence
stronger predictors of cultural variance in cognition, than these other characteristics. See Dan M.
Kahan, Donald Braman, Geoffrey L. Cohen, John Gastil & Paul Slovic,
Who Fears the HPV Vaccine,
Who Doesn't, and Why? An Experimental Study of the Mechanisms of Cultural Cognition, 34 L. & Human
Behavior 501, 505 n.5 (2010). Ideally, however, nonlinear scaling and classification techniques, such
as grade-of-membership modeling and latent class analysis, could be used to form even more dis-
cerning and hence even more predictive measures of cultural predispositions based on appropriate
combinations of cultural worldviews and related identifying characteristics. See generally Kenneth G.
Manton, Max A. Woodbury, Eric Stallard & Larry S. Corder, The Use of Grade-of-Membership Techniques
to Estimate Regression Relationships, 22 Socio. Method. 321 (1992); R. F. Potthoff, K. G. Manton & M.
A. Woodbury, Dirichlet Generalizations of Latent-Class Models, 17 J. Classification 315-353 (2000). Such
an analysis would likely add resolution and detail to the picture of motivated cognition that our data
reveal. But it would still be culturally motivated cognition that is being observed.
We used a video to elicit evidence of cultural cognition because of the utility of a visual stimulus
for the experimental design and because of the prevalence of video proof in real-world cases that
involve First Amendment challenges to restrictions on protest activity. See supra note 62. Our results
thus reinforce the concerns and cautions of authors who have emphasized the risk that judges and
jurors will invest video proofs, which are becoming increasingly common, to mislead. See Neal Fei-
genson & Christina Spiesel, Law On Display: The Digital Transformation of Legal Persuasion and
Judgment (2009);
Jessica Silbey, Cross-Examining Film, 8 U. Md. L. J. Race, Religion, Gender & Class 17 (2009);
Howard M. Wasserman, Video Evidence and Summary Judgment: The Procedure of Scott v. Harris, 91 Judicature 108
(2008). Nevertheless,
the impact of cultural cognition is by no means limited to visual perception. See,
e.g., Dan M. Kahan, Donald Braman, Paul Slovic, John Gastil & Geoff Cohen, Cultural Cognition of the
Risks and Benefits of Nanotechnology, 4 Nature Nanotechnology 87 (2009) (biased assimilation of evi-
dence in written materials). Nor is there any reason to believe that videos are a form of trial proof
uniquely vulnerable to the effects of culturally motivated cognition. See Kahan, Hoffman, & Braman,
supra note 4, at 900-01.
“They Saw a Protest” 31
B. Cognitive Illiberalism and the Constitution
The practical motivation for this study was to focus attention on the danger
that cognitive illiberalism can pose to constitutional law. We use this term to refer to
the vulnerability of political and legal decisionmakers to betray their commitment
to liberal neutrality by unconsciously fitting their perceptions of risk and related
facts to their sectarian understandings of the good life. This is the dynamic, we
believe, that transforms seemingly empirical debates over how to protect the envi-
ronment, promote public health, and secure the nation from external threats into
occasions for divisive group-based status competition.
Our study results show
how readily constitutional decisionmaking can become infected by this pathology.
In our subjects, cognitive illiberalism eviscerated the line between speech and
conduct. The “speech”-“conduct” distinction can be seen as one doctrinal device
courts employ to test whether a regulation conforms to liberal prohibition on go-
vernmental promotion of a moral or political orthodoxy: by requiring that a regu-
lation be shown to promote a governmental interest independent of hostility to
any particular idea, the teleological conception of the “speech”-“conduct” divide
assures that law is used to pursue secular goods of value to all citizens regardless
of their cultural outlooks.
Enforcing this test, however, necessarily requires deci-
sionmakers to make critical determinations of fact: in the case of a mass demon-
stration, for example, did the protestors intend to intimidate or only persuade; were
the protestors simply expressing impassioned dissent or did they impose them-
selves on members of the public in an assaultive or invasive manner (e.g., “scream-
ing in their faces”); were onlookers genuinely frightened of physical assaultor
merely angry, offended, or possibly even ashamed by exposure to the protestors’
message; did law enforcement actors intervene to preempt incitement to violence
or only to quell a public backlash propelled by animosity toward the demonstra-
tors’ point of view? For our subjects, the answers were decisively shaped by the
congruence between the protestors’ message and the subjects’ own cultural
worldviews. As a result, in the course of certifying that the law was free of cultural-
ly partisan influence, they ended up infusing it with exactly that.
Other First Amendment doctrines also seem vulnerable to this type of sub-
version. Like the speech-conduct distinction, the so-called “time, place, manner
doctrine requires that regulations be justified on the basis of an “interest unrelated
to suppression of expression.”
It thus requires the same sort of fact-finding,
subject to the same danger of motivated cognition and the same dangers of polari-
See Dan M. Kahan,
The Cognitively Illiberal State, 60 Stan. L. Rev. 115 (2007.
See Rubenfeld, supra note 13, at 818-19; supra pp. 7-8.
See Clark v. CCNV, 468 U.S. 288, 293 (1984).
Cf. id. at 299 (“standard . . . for validating a regulation of expressive conduct . . . in the last analysis
is little, if any, different from the standard applied to time, place, or manner restrictions”).
“They Saw a Protest” 32
In addition, valid “time, place, manner” restrictions must be “content neu-
tral”that is, applicable without regard to speakers’ topics or points of view.
Regulators might try to evade this requirement by resort to seemingly general regu-
lations (say, that marchers acquire liability insurance for a particular level of dam-
ages) that, in practice, meaningfully limits only a disfavored point of view (Ku
Klux Klan members proposing to march in a predominantly Jewish communi-
Or regulators might apply a facially neutral provision (e.g., on the number
of groups that will be issued permits to march on a given day) in a selective man-
ner that reflects their animosity toward a particular message or idea (“Irish gay
pride,” on St. Patrick’s Day).
We expect searching First Amendment review to
give the lie to such stratagems.
But if legal decisionmakers’ cannot shake the
influence of culturally motivated cognition, how can we be confident that they
themselves will reliably perceive that a regulator is defying the content-neutrality
requirement in one of these ways? Why shouldn’t we expect those decisionmakers
to be perceived by those who see things otherwise as having fit their conclusions to
their values?
The same dynamics inhere in First Amendment standards relating to “unpro-
tected” and “low value” categories of speech. Authorities can ban obscenity but not
pornography, fighting words but not dissent: yet cultural meanings might well be ex-
actly what a legal decisionmaker is perceiving when she determines that a particu-
lar form of speech (a depiction of a woman enjoying adulterous sexor forced
; the burning of a flag
or a cross
) is one and not the other. The First
See, e.g., See, e. g., Consolidated Edison Co. v. Public Service Comm'n, 447 U.S. 530, 535-536 (1980).
Cf. Collin v. Smith, 578 F.2d 1197, 1208 (7th Cir. 1978) (“[T]he Village has flatly prohibited First
Amendment activity, not itself directly productive of the feared injury, by those too controversial to
obtain commercial insurance.”).
Cf. Olivieri v. Ward, 637 F. Supp. 851 (S.D.N.Y. 1986) (finding “logistical” concerns to be a pre-
text for denying gay rights group a permit to assemble in public forum along St. Patrick’s Day parade
route). But cf. Irish Lesbian and Gay Organization v. Giuliani, 918 F. Supp. 732 (S.D.N.Y. 1996) (uphold-
ing reliance on content-neutral criteria involving traffic disruption and public safety to deny marching
permit to gay and lesbian group).
See generally Burson v. Freeman, 504 U.S. 191, 213 (1992) (“In some cases, a censorial justification
will not be apparent from the face of a regulation which draws distinctions based on content, and the
government will tender a plausible justification unrelated to the suppression of speech or ideas.
There the compelling-interest test may be one analytical device to detect, in an objective way, wheth-
er the asserted justification is in fact an accurate description of the purpose and effect of the law.”)
(Kennedy, J., concurring); Kagan, supra note 7, at 454 (“the strict scrutiny test operates as a measure
of governmental motive”); Rubenfeld, supra note 13, at 786 (First Amendment strict scrutiny best
conceived of as a “a device for smoking out impermissible purposes”).
Cf. American Booksellers
Ass'n, Inc. v. Hudnut, 771 F.2d 323 (7th Cir. 1985).
Cf. American Booksellers Ass'n, Inc. v. Hudnut, 771 F.2d 323 (7th Cir. 1985).
Cf. R.A.V. v. St. Paul, 505 U.S. 377, 416 (1992) (Stevens, J., concurring) (cross-burning has
“communicative content-- communicative contenta message of racial, religious, or gender hostili-
“They Saw a Protest” 33
Amendment demands proof when regulators invoke “secondary effectstraffic
congestion, disruption of commerce, increased incidence of crime, and the like
to justify zoning restrictions on strip clubs and other forms of “low value”
But if legal decisionmakers’ own ability to weigh the proffered evidence
is affected by motivated cognition, they will do a poor, or at least a suspect, job of
distinguishing pretext from truth.
Indeed, we suspect this point can be generalized to constitutional theorizing
as a whole. As discussed, the First Amendment can be integrated into a general
theory that reads the Constitution as implementing the liberal prohibition on state
endorsement of partisan conceptions of the good life.
Like the First Amend-
ment, the Equal Protection and the Due Process Clauses require courts to “strictly
scrutinize” proffered secular rationalespublic health, deterrence of criminal vi-
olence, national security, and the liketo “flush out” the impact, conscious or
unconscious, of regulators’ animosity toward those whose identity or values defy
dominant norms.
But if legal decisionmakers, like everyone else, are cognitively
motivated by their cultural affiliations, then theylike everyone elseare more or
less likely to see challenged laws as contributing to attainment of secular ends de-
pending on whether those laws affirm or denigrate their own cultural commit-
ments. Angry denunciations of judges who have thrown their lot in with one or
the another of the belligerents in the American “culture wars” is itself a form of
status conflict characteristic of cognitive illiberalism.
Some legal commentators
(and historically certain jurists
) have criti-
cized constitutional standards that “balance” constitutional liberties against “com-
Cf. United States v Eichman, 496 U.S. 310, (1990) (Stevens, J., dissenting) (“Th[is] case has nothing
to do with ‘disagreeable ideas.’ It involves disagreeable conduct. . . .”).
See generally City of Los Angeles v. Alameda Books, Inc., 535 U.S. 425, 438-39 (2002) (explaining that
governing authority must furnish “evidence . . . [to] support its rationale” and cannot “get away with
shoddy data or reasoning”).
See supra pp. 7-8
See City of Cleburne v. Cleburne Living Ctr., 473 U.S. 432, 471, (1985) (Marshall, J., concurring)
(quoting United States v. Carolene Products Co., 304 U.S. 144, n. 4 (1938)):
The discreteness and insularity warranting a ‘more searching judicial inquiry’ must . . . be
viewed from a social and cultural perspective as well as a political one. To this task judges
are well suited, for the lessons of history and experience are surely the best guide as to
when, and with respect to what interests, society is likely to stigmatize individuals as mem-
bers of an inferior caste or view them as not belonging to the community. . . .
See generally John Hart Ely, Democracy and Distrust 146 (1980) (strict scrutiny “smokes out” illicit
motivation in Equal Protection analysis); Jed Rubenfeld, Affirmative Action, 107 Yale L.J. 427, 436-37
(1997) (same).
See generally J.M. Balkin, The Constitution of Status, 106 Yale L.J. 2313 (1997).
See, e.g., Jed Rubenfeld, Affirmative Action, 107 Yale L.J. 427, 441-43 (1997) (arguing against inter-
est-balancing in Equal Protection analysis); Rubenfeld, supra note 13, at 778-93 (arguing against inter-
est-balancing in Free Speech analysis).
“They Saw a Protest” 34
pelling interests,” such as national security, public order, and diversity. The phe-
nomenon of culturally motivated cognition vindicates their anxiety that such “tests
inevitably become intertwined with the ideological predispositions of those doing
the balancing.”
But our study results suggest that these commentators are too
quick to assume that their preferred alternative to balancingsuch as the “teleo-
logical conception” of the speech-conduct distinction, the “anti-caste” prin-
the liberal “harm” criterion,
and the likewill necessarily avoid such
entanglement. The primary implication of our studythe main message we are
trying to get acrossis that constitutional theorists have paid too much attention
to explicating the normative content of various free speech standards and too little
to the psychology of enforcing them.
C. Judges, Jurors, and Citizens
We have been assessing the potential impact and implications of culturally
motivated cognition on constitutional decisionmaking. It should not be assumed,
however, that all constitutional fact finders think in the same way. The design and
sample we used in this study furnish evidence of how members of the public
might be influenced by cultural cognition as jurors called upon to make findings of
fact pertinent to the speech-conduct distinction and related doctrines.
But of-
tentimeswhen protestors seek a preliminary injunction against police interfe-
rence with a planned rally or march, for examplejudges will make these sorts of
findings. Indeed, how likely such determinations are to be made by judges rather
than jurors can be influenced by making appropriate adjustments in the standards
used to decide threshold motions or to review findings of fact on appeal. Do the
study results furnish insight on how factfinding and related decisionmaking tasks
should be allocated between judges and jurors?
See Konigsberg v. State Bar, 366 U.S. 36, 61 (1961) (Black, J., dissenting) (“I believe that the First
Amendment's unequivocal command that there shall be no abridgement of the rights of free speech
and assembly shows that the men who drafted our Bill of Rights did all the ‘balancing’ that was to be
done in this field.”).
Ely, supra note 10, at 1500.
See Cass R. Sunstein, The Anticaste Principle, 92 Mich. L. Rev. 2410 (1994)
See Lior Jacob Strahilevitz, Consent, Aesthetics, and the Boundaries of Sexual Privacy after Lawrence v.
Texas, 54 DePaul L. Rev. 671 (2005); Paul M. Secunda, Lawrence's Quintessential Millian Moment and its
Impact on the Doctrine of Unconstitutional Conditions, 50 Villanova L. Rev. 117 (2005).
Studies suggest that mock jurors’ reactions to detailed trial vignettes is strongly predictive of how
they respond to more vivid forms of proof, including the testimony of live witnesses. See Brian H.
Bornstein, The Ecological Validity of Jury Simulations: Is the Jury Still Out?, 23 Law & Human Behavior 75
(1999). In addition, the views of individual jurors after consideration of the evidence is generally
thought to be highly predictive of how they’ll vote at the conclusion of deliberations, see Dennis J.
Devine, Laura D. Clayton, Benjamin B. Dunford, Rasmy Seying & Jennifer Pryce, Jury Decision Mak-
ing: 45 Years of Empirical Research on Deliberating Groups, 7 Psychology, Public Policy, & Law 622, 690
(2001). Once the relationship between individual cultural worldviews and first-ballot votes is estab-
lished, computer simulations can furnish additional insight into the probability of final verdicts in
various kinds of cases conditional on the the worldviews (and other characteristics) of the individuals
on any particular jury. See Maggie Wittlin, The Results of Deliberation, (unpublished June 15, 2011),
available at
“They Saw a Protest” 35
We addressed a similar question in a previous article. In it, we examined the
impact of cultural cognition on perceptions of a high-speed car chase that came to
an end when the police deliberately rammed the fleeing motorist’s vehicle, causing
a fiery crash that left the driver paralyzed. The Supreme Court (in an 8-1 opinion)
had held that “no reasonable jury” whose members viewed a videotape of the
chase shot from inside the pursuing police cruisers could dispute the “factual issue
[that] respondent was driving in such fashion as to endanger human life.”
when we showed the video to a representative sample of over 1,000 members of
the public, we found significant levels of disagreement between cultural groups on
exactly that. We thus concluded that judges’ own perceptions of fact can some-
times furnish them with unreliable guidance on what “reasonable” but culturally
diverse people are likely to perceive,
a position since forcefully amplified by
critics of Supreme Court decisions that expands the power of judges to grant mo-
tions for dismissal or for summary judgment.
The results of the present study might be understood to furnish even more
support for such a critique. It reinforces our previous study with experimental evi-
dence that what people see in trial proof will often turn on who they are. If one
thinks that adjudication will be more accurate or more legitimate if informed by a
diversity of culturally grounded perceptions, then the law should be fashioned and
applied in a manner that fortifies the central role of the jury in determining the
But we can also see how, in the context of this study in particular, one might
draw exactly the opposite conclusion. After all, one might well think that the point
of constitutional review is to insulate the law (or at least certain aspects of it) from
the influence ofand conflict overpartisan worldviews. Indeed, our finding that
one and the same individual might well see the facts differently depending on her
evaluation of protestors’ messages seems to involve exactly the sort of content-
based discrimination that the First Amendment is understood to prohibit. In this
and similar types of cases, then, one might advocate enlarging the role of courts in
factfinding through a relatively aggressive exercise of summary adjudication pow-
Scott v. Harris, 550 U.S. 372 380 (2007).
See Dan M. Kahan