reflection

Journal of Applied Research in Memory and Cognition 7 (2018) 387–397

Contents lists available at ScienceDirect

Journal of Applied Research in Memory and Cognition

j ourna l h om epage: www.elsev ier .com/ locate / ja rmac

Holding Yourself Captive: Perceptions of Custody During Interviews and Interrogations

Fabiana Alcestea,b,∗, Timothy J. Lukea,c, Saul M. Kassina

a John Jay College of Criminal Justice, City University of New York, United States b The Graduate Center, City University of New York, United States

c University of Gothenburg, Sweden

Police custody activates important legal safeguards. To determine custody, courts examine objective conditions and ask whether a “reasonable person” would feel free to leave while being questioned. In Study 1, student participants were either interviewed or interrogated about a staged theft they believed to be real. Interviews and interrogations embodied specific factors considered noncustodial or custodial, respectively. Observers then watched videos of these sessions. Participants in interviews did not feel significantly freer to leave than those in interrogations, though observers did make this distinction. In Study 2, some participants were explicitly pre-advised of their freedom to leave. The advisement induced participants to report they were free to leave as an objective matter but did not significantly affect their subjective feelings of freedom. In both studies, the actor–observer divergence vanished when observers imagined themselves from the actor’s perspective. These results challenge legal assumptions about custody and suggest lines of future research.

General Audience Summary When police interrogate someone about a crime in an effort to get a confession, they must inform that person of his or her Miranda rights to remain silent and have a lawyer present. But this right comes into play only when the person being questioned is “in custody.” But what it means to be in custody, and not feel free to leave, has never been empirically examined. In this article, we report the results of two experiments in which Phase 1 laboratory participants were interviewed as witnesses or interrogated as suspects about an alleged theft that occurred in their presence; Phase 2 neutral observers then watched videotapes of these prior sessions. Overall, results showed that most Phase 1 participants—even those questioned in a non-accusatory manner, as witnesses—felt as if they were in custody and were not free to leave. In contrast, Phase 2 observers differentiated between the two types of sessions, perceiving participants as free to leave while being interviewed but not when interrogated as suspects. Interestingly, however, when observers were asked to imagine themselves in the participant’s situation, the majority reported that they would not have felt free to leave—even from the “noncustodial” interview. Also interesting is that an explicit “You are free to leave at any time” advisement did not significantly increase subjective perceptions of freedom, as assumed by law. These studies thus contradict assumptions made by U.S. courts about custody and, therefore, the protections that are supposed to be activated.

N U v

Keywords: Police custody, Interview, Interrogation, Perception of freedom

Author Note Fabiana Alceste, John Jay College of Criminal Justice, City University of

ew York, United States, The Graduate Center, City University of New York, nited States; Timothy J. Luke, John Jay College of Criminal Justice, City Uni- ersity of New York, United States, University of Gothenburg, Sweden; Saul

M. Kassin, John Jay College of Criminal Justice, City University of New York, United States.

∗ Correspondence concerning this article should be addressed to Fabiana Alceste, John Jay College of Criminal Justice, City University of New York, United States. Contact: [email protected]

NS O

S a a i a t S e r g 2 d (

p d a p U I t b a i s a

t t ( v c a S t i i A v t o p h c g m b a

s a s t a p a

o a s r f s w a t

w t p w ( ( a a I a p a o c

e t s w ( s b a t e f f c o b d (

a d s a c p w t m t

PERCEPTIO

In the landmark case of Miranda v. Arizona (1966), the U.S. upreme Court ruled for the first time that police must inform ll suspects in custody of their constitutional rights to silence nd to counsel—and that any statement taken without a know- ng, intelligent, and voluntary waiver would not be admissible t trial. Over the years, a number of researchers have questioned he protective effectiveness of Miranda (for an overview, see malarz, Scherr, & Kassin, 2016). In light of numerous DNA xonerations, many of which involved false confessions, recent eform efforts have focused on the requirement that all interro- ations be video recorded in their entirety (e.g., see Kassin et al., 010). At last count, approximately half of all states now man- ate on a statewide basis the video recording of interrogations Sullivan, 2016).

Both Miranda and video recording have in common that the rocedural requirement is triggered by “custody.” But what con- itions define custody in operational terms? What dispositional nd situational factors lead individuals questioned by police to erceive themselves as free, or not free, to leave? Over the years, .S. courts have struggled to define this all-important construct.

n Miranda, the Court defined a custodial interrogation as “ques- ioning initiated by law enforcement officers after a person has een taken into custody or otherwise deprived of his freedom of ction in any significant way” (p. 445). Elsewhere in that opin- on, the Court cited several indicia of a custodial interrogation uch as intimidation, trickery, a restriction of personal liberty, n unfamiliar environment, and hostility from law enforcement.

Over the years, the courts have sought to create an objec- ive test by which judges would determine custody. It is clear hat formal arrest triggers custody and all subsequent protections Orozco v. Texas, 1969). Often, however, police question indi- iduals who have not been arrested. In these more ambiguous ases, the situation may be considered custodial if police restrict n individual’s freedom of action in a significant way. Hence, in tansbury v. California (1994), the U.S. Supreme Court ruled

hat custody depends on “the objective circumstances of the nterrogation, not on the subjective views harbored by either the nterrogating officers or the person being questioned” (p. 323). lthough no comprehensive list of factors exists, the courts have ariously cited as relevant whether police informed the suspect hat he or she was free to leave, whether the suspect’s freedom f movement was restrained (e.g., was the suspect at home, in ublic, or in a police station; was he or she in handcuffs and eld in an open or locked room; were his or her shoes, clothing, ell phone, or car keys taken?), and whether coercive interro- ation techniques were used (e.g., who initiated contact; how any police officers were present; were friends or family mem-

ers present; how long did the session last; did police make ccusations and threaten physical force?).

In considering these questions, it is interesting that the Court taked out a behaviorist stimulus-response position by which

state of freedom or custody—and the decision to leave or tay—are determined by strictly objective parameters, not by

he individual’s cognitive representation of the situation and nticipated consequences of a particular response (D. Reisberg, ersonal communication, August 2, 2017). Indeed, the Court sserted that it will not defer to an individual suspect’s or police

R o t t

F CUSTODY 388

fficer’s reported perceptions. Instead it advocated a “reason- ble person standard,” the central question being whether the ituation would lead a reasonable person to feel a significant estriction on his or her freedom of action (Stansbury v. Cali- ornia, 1994; for historical origins of the “reasonable person” tandard, dating back to Adolphe Quetelet’s nineteenth century ritings, see Beirne, 1987). Put another way, “Would a reason-

ble person. . .have felt free simply to get up and walk out of he. . .room. . .at will?” (Yarborough v. Alvarado, 2005, p. 670).

Among the objective criteria that a “reasonable person” ould consider, the context and manner in which police ques-

ion a suspect looms large. During a criminal investigation, olice conduct interviews and interrogations. Within the frame- ork of the Reid Technique, first published by Inbau and Reid

1962), cited by the Miranda court, and now in its fifth edition Inbau, Reid, Buckley, & Jayne, 2013), an interview is a non- ccusatory form of questioning designed to gather information nd determine if the individual is being truthful or deceptive. f the latter judgment is made, that individual is identified as

suspect and subjected to interrogation—an accusatory, guilt- resumptive process of influence in which trickery and deception re used to elicit a confession. Although these processes may verlap, the distinction serves as a proxy for the factors that onstitute custodial and noncustodial questioning.

Despite the pivotal nature of this psychological construct, no mpirical research has examined people’s perceptions of cus- ody. Of direct relevance to this inquiry, however, are classic tudies of actor–observer differences in attribution, particularly ith regard to attributions of freedom. Beginning with Heider’s

1958) Gestalt-inspired hypothesis that “behavior. . .has such alient properties that it tends to engulf the field” (p. 54), attri- ution theorists have found that observers tend to focus on the ctor and overlook contextual factors. As a result of this atten- ional bias, people routinely commit the fundamental attribution rror, or correspondence bias, making dispositional attributions or others’ behavior while underestimating the role of situational actors (Gilbert & Malone, 1995; Jones, 1990; Ross, 1977). In ontrast, studies of self-perception indicated that actors focus utward, on aspects of the environment that impinge on their ehavior, which leads them to make situational attributions. This ivergence in perceptions is known as the actor–observer effect Jones & Nisbett, 1972; Watson, 1982).

Harvey, Harris, and Barnes (1975) examined the ctor–observer effect with specific regard to perceptions of free- om. Their study utilized a “Milgramesque” teacher–learner hock paradigm to test how participants in different roles ttribute responsibility and freedom for actions that produce onsequences of varying severity. In each session, two partici- ants were randomly assigned to the teacher or observer role, hile a confederate, ostensibly in an adjacent room, played

he learner. By random assignment, the learner exhibited either oderate or severe distress. At the end of each session, the

eacher and observer answered questions about the experience. esults showed that when the learner exhibited more distress,

bservers attributed more freedom and responsibility to the eacher who, in turn, attributed less freedom and responsibility o themselves.

NS O

b o t p a i n s i m g s t i w 2 t a i i

S o a a p c b 2 r w s t

P

u t o g A (

w w p t d p t W

w i t

w t w m a n p a a s p

w i a 3 f w u t a i l

r e d a m e t

t t t m u t r r r b t c

r i d

PERCEPTIO

The Present Studies

Two experiments were designed to assess what it means to e “in custody” from the perspective of both an actor and an bserver. Specifically, we sought to test two assumptions: that he reasonable-person standard serves as a valid proxy for a sus- ect’s perceptions of freedom and custody, and that instructing

suspect that he or she is free to leave is sufficient to elim- nate the perception of custody. In Study 1, we developed a ovel paradigm involving a staged theft to test the hypothe- is that people feel free to terminate questioning while being nterviewed as a witness (an information-gathering session that

imics a noncustodial situation) but not while being interro- ated as a suspect (an accusatory session that mimics a custodial ituation). All sessions were video recorded, after which we ested whether actors and later observers perceived the custody nquiry as the courts would predict according to condition—and hether they agreed or diverged in their perceptions. In Study , we replicated the non-accusatory interview paradigm to test he Supreme Court’s hypothesis that participant actors randomly ssigned to receive an explicit “free to leave at any time” pre- nstruction would feel freer to terminate than those not similarly nstructed.

Study 1

As a result of extensive pilot testing involving 40 sessions, tudy 1 was conducted in two distinct phases. Phase 1 consisted f a live laboratory experiment involving interactions between n experimenter, confederate, research assistant, security guard, nd participant. Our objective was to determine whether partici- ants in live questioning sessions, led to believe that a theft was ommitted and that the stakes were real, felt free to leave while eing interviewed versus being interrogated by the guard. Phase

consisted of an online study in which observers watched and esponded to videotaped sessions from Phase 1. Each observer atched a single session after which he or she was asked the

ame questions put to the actors concerning their freedom to erminate questioning and leave.

hase 1 Method

Participants. Sixty undergraduate psychology students vol- nteered to participate as part of their course requirement. Within his sample, 61.7% were female; all were between the ages f 18 and 35 (M = 20 years, SD = 2.71). In terms of demo- raphic makeup, 45% self-identified as Hispanic; others were sian (20%), Black (15%), White (11.7%), and other ethnicities

6.7%). Students who volunteered were led to believe that the purpose

as to examine perceptions of autonomy. To screen for those ho might have an adverse reaction to our inherently stressful rocedure, we asked all prospective participants, prior to arrival, o indicate whether they had ever been diagnosed with an anxiety

isorder (options: Yes, No, Prefer not to respond) and to com- lete the Beck Anxiety Inventory (BAI), which asks respondents o self-report recent experiences of various symptoms (Beck,

ard, Mendelson, Mock, & Erbaugh, 1961). Fifteen students

a t t w

F CUSTODY 389

ho answered “Yes” or “Prefer not to respond” to the anx- ety disorder question or who scored above the high-anxiety hreshold of the BAI (>36 points) were excluded.

Stage 1: the missing wallet. Upon their arrival, students ere seated in a small waiting area with a rectangular table and

wo chairs. The front door was open to the adjacent hallway, here students assembled between classes. A female experi- enter explained that sessions were running behind schedule

nd the prior participant was still working on a final question- aire. This prior participant was actually a confederate, soon to lay the role of theft victim. The participant sat in the waiting rea until the confederate was finished. The experimenter then sked to see a college identification card to verify his or her ign-up name; this was included to bolster the credibility of the rocedure that would follow.

The participant read and signed the informed consent, after hich the experimenter re-joined the confederate in the exper-

mental room, leaving the participant alone in the waiting area longside the confederate’s purse and an open door. After 2 or

min, the experimenter and confederate came out. The con- ederate, gathering her belongings to leave, stated that her allet—where she had kept the college I.D. card she had earlier sed—was missing from her purse. After searching frantically, he confederate stated that she had to leave for a scheduled ppointment. The experimenter advised her to report the miss- ng wallet to Public Safety, which she said she would do before eaving.

Stage 2: the cover story. After the confederate left, a female esearch assistant entered the waiting area. The experimenter xplained that this research assistant was a fellow graduate stu- ent who was conducting another experiment. The experimenter sked the participant if he or she was willing to take a few inutes to help her out before starting the original study. The

xperimenter noted that it was not possible to receive credit for his favor. Still, all participants acceded to the request.

This “detour” was built into the protocol to instill in par- icipants an incentive to hasten their upcoming sessions with he security guard so they can move on to the study for which hey had registered. The added time pressure was designed to

otivate participants to exercise their freedom to escape the sit- ation. Hence, we anticipated that some number would attempt o terminate their interviews or interrogations because of their esponsibility to the experimenter and desire to earn credits. The esearch assistant and participant then entered the experiment oom, which was modeled after an interrogation room: small, arely furnished, windowless, and furnished with a rectangular able with three chairs. At that point, the participant was handed over-story questionnaires about perceptions of autonomy.

Stage 3: the questioning session. After 5 min, a male esearch assistant knocked on the experiment room door. Prohib- ted by law from impersonating a police officer, this confederate, ressed in civilian clothing, entered the room with an air of

uthority and equipped with a radio and key ring visibly attached o his belt. He introduced himself as a security guard and said hat he was investigating a missing wallet report. He said he anted to question the participant.

NS O

t g r r t q i I fi R c t t r

l m w d t q g r s t t

m t s E i o T w a “ t m c

i t o f a ( t a a t w t

i p

a t t k r b F t t e

P

o o v t B

U c i f t 2 o l 1 w ( y

c a c t i f t d

w p a w e 2 u l

t P

PERCEPTIO

Prior to each session, participants were randomly assigned o an interview or interrogation condition. In the interview, the uard questioned the participant with the door open and the esearch assistant present. In the interrogation, he asked the esearch assistant to step outside and shut the door. Although hese interactive sessions could not be scripted verbatim, the uestioning followed the training and manual guidelines for nterviews or interrogations, respectively, as recommended by nbau et al. (2013). Indeed, the protocols were designed by the rst author, who had completed training and is certified in the eid Technique. A hidden surveillance camera mounted in the orner of the ceiling captured an equal-focus view of the par- icipant and guard. This camera started to record as soon as he guard entered the room. Participants were neither physically estrained nor told that they could not leave.

Interviews. In this condition, the guard adhered to the guide- ines in the Reid Technique by asking non-accusatory questions

odeled after the Behavioral Analysis Interview. All interviews ere conducted with the research assistant in the room and the oor wide open. This interview consisted of open-ended ques- ions (e.g., “Can you tell me what happened here?”) and direct uestions (e.g., “Let me ask you straight out: did you take this irl’s wallet?”), as well as specific behavior-provoking questions ecommended in the Reid Technique (e.g., “What do you think hould happen to the person who took the wallet?”). The guard erminated the interview by thanking the participant and asking he research assistant if he could speak to her outside.

Interrogations. In this condition, participants underwent a odified version of the Reid Technique of interrogation. Before

he session, the guard asked the research assistant to step out- ide to talk. Moments later, he re-entered alone and shut the door. ach session began with a short interview to set the context. The

nterrogation itself then opened with a direct accusation (“Based n what I’ve seen and heard, I know you took this girl’s wallet. hat much is clear. What I’m trying to figure out is why”), at hich point the guard introduced standard minimizing themes

nd the notion that the theft was a spur-of-the-moment act (e.g., I don’t believe you planned to take this girl’s wallet”). All par- icipant objections and denials were interrupted. After several

inutes, the guard terminated the session and left, he said, to all his superior.

Questionnaire and debriefing. After each session, the exper- menter returned and partially debriefed the participant by noting hat the purpose of the experiment was not to study perceptions f autonomy, as previously stated, and that she would explain urther in a few moments. Participants were then administered

perceptions of custody questionnaire in which they indicated a) whether they were free to leave the room while being ques- ioned, and (b) how free they felt they were to leave (they were lso then asked to indicate whether and how often they thought bout leaving, how motivated to leave they were, how motivated o stay they were, whether they had asked permission to leave, hether they physically tried to leave, and why they stayed in

he room). Serving as a manipulation check on the effectiveness of the

nterview versus interrogation manipulation, we asked partici- ants to indicate whether, in their view, they were questioned as

q “ t c

F CUSTODY 390

witness to the theft or as a suspect, whether the guard believed hey were guilty or innocent, and their degree of confidence in his latter belief. Participants were also asked to indicate if they new what it means, in legal terms, to be “in custody,” and to ate the extent to which they felt like they were in custody while eing questioned. All ratings were made on a 1–10 point scale. inally, participants answered open-ended questions about why

hey chose to stay in the room during questioning and what hey thought of the experiment. Once this was completed, the xperimenter fully debriefed the participant.

hase 2 Method

Phase 2 aimed to assess observers’ perceptions of freedom verall and between the two questioning conditions. This part f the study was conducted as a Qualtrics survey and distributed ia Amazon’s Mechanical Turk, an online platform that enables he rapid and inexpensive collection of high-quality data (see uhrmester, Kwang, & Gosling, 2011).

Participants. Phase 2 involved 303 participants from the .S., 50 of whom were automatically excluded for failing to

omply with instructions (i.e., they took notes, searched the nternet or completed other tasks during the study, asked others or help, refreshed the webpage, or extended their participa- ion into a second session). The final sample thus consisted of 53 participants, each paid $0.50 for their time. Overall, 58% f our sample was female and 42% was male; 48.5% had at east a Bachelor’s degree, 40% had completed some college, and 1.5% had a high school education. A majority of the sample as White (73.5%), followed by Black (9.1%), Hispanic/Latino

9.1%), Asian (6.3%), and other (2%). The mean age was 38.6 ears old (SD = 13.02).

Materials and procedure. After providing online informed onsent, MTurk workers were given contextualizing information bout the videos from Phase 1. They were told that a study was onducted at a city university, that a wallet went missing, and hat a Public Safety officer questioned a participant as part of his nvestigation. Each MTurk worker then watched one of 57 videos rom Phase 1 (two Phase 1 participants declined consent to use heir videotape; one session failed to record due to technical ifficulties).

Of the 253 observers, 142 watched an interview and 111 atched an interrogation. They were then asked about their erceptions of the person’s guilt, whether that person was

witness or suspect, whether he or she was free to leave, hether he or she felt free to leave, and why—in an open-

nded question—the person stayed during questioning. Phase observers were then asked to imagine themselves in the sit- ation and indicate whether they would personally feel free to eave.

Coding. In both phases, two independent raters—blind as to he purpose of the study and hypotheses—coded Phase 1 and hase 2 participants’ open-ended responses to their respective

uestions, “Please describe why you stayed in the room” and Why do you think the person stayed in the room?” Based on he authors’ preliminary assessments, raters were given a priori ategories and instructed to record the presence or absence of

NS O

a d A x

P

d c ( q t

t d w g c t S 2 t t i w t t N t A f s c N h

m r w O ( g w r 1

a w d t o I o m N t

f d t i t t

r a d h b i t q a t v v d m S d

t O t d h w i g t n n g o N t p

P

c s o o

w s ( t

PERCEPTIO

ll category instances per question. The raters each coded a ran- om sample of 20% of all open-ended responses for each phase. fter achieving an acceptable level of inter-rater reliability (all

s > .67), each coder rated half of all questions.

hase 1 Results

Phase 1 interviews and interrogations were designed to istinguish between two important types of questioning—non- ustodial and custodial. A total of 60 sessions were conducted 29 interviews and 31 interrogations). We refer to participants uestioned in Phase 1 as “actors” and MTurk workers who saw he videotapes in Phase 2 as “observers.”

Manipulation checks. Three variables served as manipula- ion checks to ensure that the interviews and interrogations were istinguishable: session length, participants’ beliefs that they ere treated as a witness or a suspect, and their beliefs that the uard perceived them to be innocent or guilty. A comparison of onditions indicated that interviews were significantly shorter han interrogations (M = 443.71 s, SD = 80.37 vs. M = 591.06, D = 118.79), t(57) = 5.52, p < .001, d = 1.44, 95% CI [0.87, .01]. Participants were asked to indicate whether, in their view, hey were questioned as a witness to the theft or as a suspect. In he interrogation condition, 30 of 31 actors (97%) reported feel- ng like a suspect. In the interview condition, only 12 of 28 actors ho answered the question (43%) felt like a suspect. Confirming

hat the interrogation was experienced as more guilt-presumptive han the interview, this difference was highly significant, χ2 (1,

= 59) = 20.85, p < .001. Participants were also asked whether hey thought the guard believed them to be innocent or guilty. gain confirming the manipulation, actors—all of whom were

actually innocent—were more likely to believe that the guard aw them as guilty in the interrogation than in the interview ondition (100%, n = 31 vs. 41%, n = 29, respectively), χ2 (1,

= 60) = 25.36, p < .001. Across conditions, these beliefs were eld with high levels of confidence (M = 7.67, SD = 2.71).

Finally, participants were asked, “Do you know what it eans, in legal terms, to be in custody?” Overall, 49 (83%)

esponded in the affirmative. Next they rated the extent to hich they felt like they were in custody during questioning. n a 1–10 point scale, the overall mean was at the midpoint

M = 4.85, SD = 3.42). Importantly, those who were interro- ated believed that they were in custody more than those who ere interviewed (M = 5.84, SD = 3.49 vs. M = 3.79, SD = 3.06,

espectively), t(58) = 2.41, p = .019, d = 0.62, 95% CI [0.10, .14].

Main dependent measures. After each session, actors nswered two main questions concerning their experience. First, e asked them to indicate if they were objectively free to leave uring the questioning (“Were you free to leave the room while he security guard was questioning you?”—yes or no). Overall, nly 19 participants (31.7%) said that they were free to leave. nterestingly, the perception of non-freedom was the norm not

nly in the accusatory interrogation condition but also in the ore neutral interview (26% and 38%, respectively), χ2 (1,

= 60) = 1.02, p = .31. Next we asked a more subjective ques- ion, “How free to leave did you feel?” On a scale from 1 (did not

t ( s v

F CUSTODY 391

eel free at all) to 10 (felt totally free), the mean rating across con- itions was well below the midpoint at 3.67 (SD = 3.13). As with he binary judgments, these ratings were not significantly higher n response to neutral interviews than accusatory interroga- ions (M = 4.17, SD = 3.35 vs. M = 3.19, SD = 2.88; respectively), (58) = 1.22, p = .23, d = 0.31, 95% CI [−0.20, 0.82].

On the question of whether they had considered leaving the oom during questioning, only 4 of 59 actors said they thought bout leaving; there were no differences as a function of con- ition, χ2 (1, N = 59) = .87, p = .35. When asked how often they ad thought about leaving on a 10-point scale, the difference etween interview and interrogation conditions was not signif- cant (M = 1.25, SD = .64 vs. M = 2.2, SD = 2.61, respectively), (21.27) = 1.58, p = .128, d = 0.5, 95% CI [−0.13, 1.13]. On the uestion of how much they wanted to leave the room, the over- ll mean was quite low (M = 1.77, SD = 1.96), though actors in he interrogation condition reported a significantly greater moti- ation than those who were interviewed (M = 2.36, SD = 2.63 s. M = 1.18, SD = 0.48, respectively), t(28.77) = 2.34, p = .027,

= 0.62, 95% CI [0.09, 1.16]. On the converse question of how uch they wanted to stay, overall ratings were high (M = 8.45,

D = 2.12); there were no significant differences between con- itions, t(58) = .25, p = .81, d = 0.06, 95% CI [−0.44, 0.57].

Finally, as described earlier, we coded open-ended responses o the prompt “Please describe why you stayed in the room.” verall, the most prevalent explanations participants cited were

hat they were innocent and had nothing to hide (“I knew that I id not take anything, so I was perfectly fine”; “I had nothing to ide, so I wanted to stay and clarify everything”), that leaving ould appear suspicious (e.g., “If I had left, that could land me

n more trouble”), and that they wanted to help in the investi- ation (e.g., “to provide helpful insight to the person who lost heir wallet”). Miscellaneous low-frequency responses included eeding to stay to earn class credit; wanting to be respectful, ot rude; and believing it was the guard’s prerogative to investi- ate. Interestingly, participants cited their actual innocence more ften in the interrogation condition (48.4% vs. 20.7%), χ2 (1,

= 60) = 5.05, p = .025; they cited wanting to help more often in he interview condition (37.9% vs. 16%), χ2 (1, N = 60) = 3.64,

= .056.

hase 2 Results

Manipulation checks. Paralleling Phase 1 manipulation hecks, we asked observers to estimate the length of the ses- ion, indicate whether the suspect was questioned as a witness r a suspect, and indicate whether they saw the actor as guilty r innocent of the theft.

Regarding the length of questioning, observers who atched interrogations accurately estimated the session as

ignificantly longer than those who watched interviews M = 618.6 s, SD = 311.5 vs. M = 434.3 s, SD = 250.1 respec- ively), t(251) = 5.22, p < .001, d = 0.66, 95% CI [0.41, 0.92]. On

he question of how the Phase 1 actor was treated, 176 observers 70%) viewed their Phase 1 participant as a suspect—more o among those who watched an interrogation than an inter- iew (86.5% vs. 56.3%, respectively), χ2 (1, N = 253) = 26.75,

NS O

p i a b p s w S C

o w t a t i N l c t g 9

a h s M i i c

1 i f g c W d

r r t n

i d p t i a a p

p t a t t 2 c o i w s i

P

e w t a ( s (

a s f p

T P

PERCEPTIO

< .001. Next, observers were asked about the actor’s guilt or nnocence. Across conditions, 31% (n = 78) saw the innocent ctor as guilty. These perceptions did not differ significantly etween conditions, χ2 (1, N = 252) = .083, p = .77. On a 10- oint scale, however, those who watched an interrogation were ignificantly more confident in their judgments than those ho watched an interview (M = 6.41, SD = 2.08 vs. M = 5.81, D = 2.16, respectively), t(250) = 2.22, p = .027, d = 0.28. 95% I [0.03, 0.53].

Main dependent measures. On the key binary question f whether the actor was free to leave the room, responses ere evenly divided at 50% across conditions (N = 126). Unlike

he actors, observers differed significantly on this judgment as function of condition: although only 35% (n = 39) believed hat the actor was free to leave the interrogation, that number ncreased significantly to 62% (n = 87) in the interview, χ2 (1,

= 252) = 17.53, p < .001. On the 10-point rating of how free to eave the actor felt, observers continued to distinguish between onditions. Specifically, they saw the actors as significantly freer o leave during interviews (M = 3.89, SD = 2.55) than interro- ations (M = 2.79, SD = 1.93), t(251) = 3.78, p < .001, d = 0.48, 5% CI [0.23, 0.73].

Finally, we asked observers to imagine themselves in the ctor’s position and to indicate whether they personally would ave felt free to leave. Overall, only 33% (n = 83) of all observers aid that they would feel free to leave in the situation they saw.

irroring the results obtained from the actors themselves, strik- ngly, this imagined sense of captivity was found not only in the nterrogation condition (30%, n = 33) but also in the interview ondition (35%, n = 50), χ2 (1, N = 252) = .762, p = .38.

To sum up, in a personally consequential situation, Phase participants clearly distinguished between the interview and nterrogation, the latter lasting longer in duration, making them eel like suspects, and making them feel they were presumed uilty. Those who were interrogated believed that they were “in ustody” to a greater extent than those who were interviewed.

hile judging most participants as innocent, Phase 2 observers iscerned these differences as well.

With our key manipulation in place, three important sets of

esults were obtained (see Table 1). First, most Phase 1 actors eported that they were not, objectively speaking, free to leave; hey also did not feel free to leave. This pattern was found ot only in the interrogation condition but surprisingly in the

i b w f

able 1 articipants’ Reactions to Interviews Versus Interrogations in Study 1

Measures Interview Interrogation

Actors Were you free? (yes/no) 38% 26%

Observers Was actor free? (yes/no) 62% 35%

Would you feel free? (yes/no) 35% 30%

Actors How free did you feel? (1–10) 4.17 3.19

Observers How free did he/she feel? (1–10) 3.89 2.79

F CUSTODY 392

nterview condition as well. Second, Phase 2 observers not only istinguished between interviews and interrogations; they also erceived the psychological implications of custody, indicating hat the actors were freer to leave, and felt freer to leave, during nterviews. Third, however, while observers saw Phase 1 actors s free to leave the interview, they did not imagine themselves s feeling free in that same situation when they took the actor’s erspective.

Study 2

Within the interview condition, which variously elicited erceptions of both freedom and custody, we used the same wo-phased paradigm to examine a bedrock Supreme Court ssumption: that explicitly advising suspects that they are free o leave and not under arrest is sufficient to obviate percep- ions of custody (California v. Beheler, 1983; Howes v. Fields, 012; Oregon v. Mathiason, 1977). While this seems a logi- al inference, Study 2 presents the first known empirical test f this assumption. We predicted that if Phase 1 actors were nstructed that they were free to leave before an interview, they ould acknowledge this on both objective and subjective mea-

ures. We further expected Phase 2 observers to be influenced n a similar manner.

hase 1 Method

Participants. Thirty-six students participated in Phase 1 in xchange for $12. Of the 36 interviews conducted, 18 opened ith an explicit advisement to the participant that he or she is free

o leave during questioning; 18 interviews did not contain this dvisement. All participants were between the ages of 18 and 35 M = 22.33, SD = 4.31); 77.8% were female. Almost half of this ample self-identified as Hispanic (47.2%), followed by White 16.7%), Black (8.3%), Asian (13.9%), and other (13.9%).

Procedure. All participants were subjected to non- ccusatory interviews. We also simplified the elaborate cover tory previously used involving a research assistant seeking help or an additional survey (despite this detour, Study 1 partici- ants reported low motivation to leave). Each session, therefore,

nvolved an experimenter, confederate, and security guard. As efore, each session opened with the discovery of a missing allet, after which participants were taken to a small, sparsely

urnished room to complete a series of questionnaires.

p-value Effect size 95% CI for effect size

.31 ϕ = 0.13 [−0.25, 0.77]

<.001 ϕ = 0.26 [0.28, 0.79] .38 ϕ = 0.055 [−0.14, 0.36]

.23 d = 0.31 [−0.20, 0.82]

<.001 d = 0.48 [0.23, 0.73]

NS O

r t l m T m k S b c A a s i o q a a l w T i a

P

v v

S i o h r w ( y

M t p P w w 1

e f t t w a i h a

A r

P

a v A p p l ( p t n p

f m d p w s a t N a i a i o [ 0 p t p

P

( fi c i ( i t g F n s (

PERCEPTIO

Before each session, participants were randomly assigned to eceive or not receive an advisement from the security guard hat they were free to leave (“Just so you know, you’re free to eave. I’m not holding you here. If you don’t want to talk to e or need to leave for any reason, you’re free to do that”). his constituted the only difference between conditions. Five inutes into the cover-story questionnaires, the security guard

nocked and entered the room. He introduced himself as a Public afety officer investigating a missing wallet and seated himself etween the experimenter and participant. All sessions were onducted with the door open and the experimenter present. s before, interviews consisted of open-ended and direct non-

ccusatory questions. After 5 min, the guard terminated the ession by asking the experimenter to step outside. Before leav- ng, the experimenter handed participants the final questionnaire n which they indicated whether they were free to leave during uestioning, how free they felt to leave, whether they thought bout leaving, to what extent they wanted to leave, whether they sked permission to leave, and whether they physically tried to eave. We also asked participants to explain in their own words hy they stayed and what they thought about the experiment. hey also answered the same manipulation check questions used

n Study 1. Afterward, participants were fully debriefed, paid, nd thanked for their time.

hase 2 Method

Phase 2 was conducted as a Qualtrics survey and distributed ia Amazon’s Mechanical Turk. Each observer watched one ideo from Phase 1 and indicated their perceptions of custody.

Participants. A total of 302 participants from the United tates completed this study; 52 who failed to comply with

nstructions were excluded from analyses. In the final sample f 250 participants, 40% were male, 60% were female; 53.6% ad at least a Bachelor’s degree, 35.6% had some college expe- ience, and 11% had a high school degree. Most participants ere White (75.2%), followed by Black (9.6%), Hispanic/Latino

6%), Asian (5.2%), and other (4%). The mean age was 40.27 ears (SD = 12.74).

Materials and procedure. After giving informed consent, Turk participants read a brief description to provide context for

he video. They were told that a wallet had gone missing during a sychology experiment and that they would watch a university ublic Safety officer questioning a participant. Each observer atched one of the 36 videos from Phase 1—128 observers atched a video containing the explicit reminder of freedom; 22 watched a video without this advisement.

Afterward, observers indicated their perceptions of the actor’s xperience. As in Study 1, they judged whether the actor was ree to leave, to what extent the actor felt free to leave, whether hey would feel free to leave if they were in the actor’s posi- ion, whether they thought the actor was innocent or guilty, and hether the actor was questioned as a witness or a suspect. We

lso asked observers to speculate as to why the actor stayed n the room. Two independent raters, blind to condition and ypotheses, coded these open-ended responses. They individu- lly coded a randomly selected 20% of the sample of responses.

n t m p

F CUSTODY 393

fter achieving an acceptable level of IRR (all xs > .67), each ater each coded half of the sample.

hase 1 Results

The overall average duration of the interviews was 6 min nd 15 s (M = 375.6 s, SD = 65.08 s). The No Advisement inter- iews (M = 405.7, SD = 62.85) were 1-min longer than the dvisement interviews (M = 345.4, SD = 53.37), t(34) = 3.10,

= .004, d = 1.034, 95% CI [0.34, 1.73]. Participants who were re-instructed that they were free to leave reported feeling ike witnesses significantly more often than those not advised 78%, n = 14 vs. 33%, n = 6, respectively), χ2 (1, N = 36) = 7.20,

= .007. Both groups were evenly split as to whether the guard hought they were guilty (44.4% vs. 50% in the advisement and on-advisement conditions, respectively), χ2 (1, N = 36) = .11,

= .738. As in Study 1, participants indicated whether they were

ree to leave. As predicted, those who received the advise- ent were significantly more likely to say yes than those who

id not (100%, n = 18 vs. 50%, n = 9), χ2 (1, N = 36) = 12.0, < .001. On the subjective rating of how free they felt they ere, participants explicitly advised exhibited a slight but non-

ignificant tendency to feel freer than those who were not dvised (M = 6.17, SD = 3.60 vs. M = 4.67, SD = 3.05, respec- ively), t(34) = 1.35, p = .186, d = 0.45, 95% CI [−0.21, 1.11]. ext participants rated the extent to which they thought about

nd wanted to leave the room during questioning. Overall rat- ngs on both questions were quite low (M = 1.81, SD = 1.95 nd M = 1.94, SD = 2.04, respectively). There were no signif- cant differences between advisement and control conditions n either measure, t(25.22) = 1.11, p = .276, d = 0.37, 95% CI −0.28, 1.03]; t(34) = .322, p = .749, d = 0.11, 95% CI [−0.55, .76], respectively. As in Study 1, not a single participant asked ermission or physically attempted to leave. When asked why hey remained in the room during questioning, the majority of articipants reported wanting to help.

hase 2 Results

Each Phase 2 participant watched one of the 36 interviews n = 128 free-to-leave advisement; n = 122 control). As in the rst study, most observers judged the actor to be innocent, per- eptions that did not differ as a function of condition (78% n the advisement group vs. 72% in the control group), χ2

1, N = 248) = 1.21, p = .271. There were also no differences n their confidence in these judgments (M = 6.84, SD = 2.22 in he advisement condition vs. M = 6.34, SD = 2.17 in the control roup), t(248) = 1.77, p = .078, d = 0.22, 95% CI [−0.02, 0.47]. inally, although all Phase 1 participants were subjected to a on-accusatory interview, 60% of observers saw their actor as a uspect. This perception did not vary according to condition, χ2

1, N = 250) = .11, p = .74. On the key question of whether the actor was free to termi-

ate, 80% (n = 201) of observers said yes. On this measure, the wo groups differed significantly (89% vs. 71.3% in the advise-

ent and control groups, respectively), χ2 (1, N = 250) = 12.49, < .001. This difference mirrors the result found for Phase 1

NS O

a t w t T a a S C s f y g (

t s n t f n b f Y t a a

c c o c t

p t i ( w w t

l j S f c

t w i i a a ( o a p

t h a o f

e a m l p w a a o r w

p S m t s

T P

PERCEPTIO

ctors answering the same question. On the subjective rating of he extent to which the actor felt free to leave, overall means ere at the midpoint (M = 5.12, SD = 2.79). Unlike the actors

hemselves, observers were highly responsive to the advisement. hose who watched an advisement interview assumed that the ctors felt significantly freer to leave than those who watched

control group interview (M = 5.80, SD = 2.80 vs. M = 4.54, D = 2.63, respectively), t(247) = 3.65, p < .001, d = 0.46, 95% I [0.21, 0.72]. Next observers were asked to imagine them-

elves in the actor’s position and indicate whether they would feel ree to leave. Overall, only 49% (n = 122) of all observers said es. This tendency was not significantly higher in the advisement roup than in the control group—54% vs. 44%, respectively, χ2

1, N = 249) = 2.54, p = .11. To sum up, Study 2 informed some suspects but not others

hat they were free to leave, an advisement that the courts deem ufficient to obviate perceptions of custody. This instruction did ot have the presumed effect. Actors in the advisement condi- ion took the information into account when asked if they were ree to leave. Yet these same actors did not feel freer than those ot informed. In contrast, observers who heard the advisement elieved that participants were free to leave (and also that they elt free) relative to those in the no-advisement control group. et, as in Study 1, many observers did not attribute subjec-

ive freedom themselves even in the advisement condition when sked to imagine themselves in the same situation. These results re summarized in Table 2.

Discussion

The present studies tested three assumptions, namely that: (1) ertain objective conditions will elicit differing perceptions of ustody versus freedom; (2) the custody inquiry can be evaluated bjectively using a “reasonable person” standard; and (3) non- ustodial freedom can be ensured by explicitly advising suspects hat they are free to leave.

None of these assumptions was substantiated. Most partici- ants in live sessions did not feel free to leave, regardless of the one or duration of the questioning—even when it was a brief nterview (not an interrogation) conducted by a civilian guard

not a detective) on a university campus (not in a police station) ith another person present and with the door left open, all of hich made the participant feel more like an innocent witness

han a guilty suspect. Perhaps most striking in this regard was the

l t i

able 2 articipants’ Reactions After Interviews With Versus Without Advisement in Study 2

Measures No advisement Advisement

Actors Were you free? 50% 100%

Observers Was actor free? 71% 89%

Would you feel free? 44% 54%

Actors How free did you feel? 4.67 6.17

Observers How free did he/she feel? 4.54 5.80

F CUSTODY 394

ack of psychological effect that explicit advisement had on sub- ective feelings of freedom. Although more research is needed, tudy 2 further suggests that advising a suspect of his or her reedom may not provide the resounding signal of safety that ourts believe.

At this point one can only speculate about why Phase 1 par- icipants felt so constrained. Given the quasi-legal situation they ere in, participants reported that they stayed to assist in the

nvestigation, to prove their innocence, and to appear forthcom- ng rather than suspicious. On a more general level, they may lso have reacted in compliance with a simple behavioral script,

conceptual representation of a stereotyped event sequence Abelson, 1981) such as Grice’s (1989) cooperative principle f conversational implicature that “other things being equal,

transaction should continue in appropriate style unless both arties are agreeable that it should terminate” (p. 29).

Whatever the underlying mechanism, the results suggest that he U.S. Supreme Court should revisit the “objective” factors it as identified for the custody–noncustody inquiry. Just because

person in a criminal investigation is not physically restrained r otherwise obstructed does not mean that he or she reasonably eels free to terminate questioning and to leave.

Across two studies, results also uncovered an actor–observer ffect that calls into question the invariance of the “reason- ble person” standard. Questioned about a purported theft, ost Phase 1 participants in both studies did not feel free to

eave—even when treated like witnesses and not under sus- icion. However, uninvolved observers of the latter sessions, ho watched the questioning online, believed that these same

ctors were objectively and subjectively free to leave. This ctor–observer difference casts serious doubt on the ability of bservers—police, judges, juries, and appeals courts—to cor- ectly judge the state of mind of suspects who are in custody hen estimating their subjective freedom. The one unanticipated exception to this divergence in

erceptions of freedom occurred when observers in both tudies 1 and 2 imagined themselves in the actor’s predica- ent. While attributing subjective freedom to the actors in

he interview condition—a perception that the actors did not hare—observers consistently reported that they themselves,

ike the actors, would not feel free to leave. On a measure hat asked them to adopt the actor’s perspective, the difference n perceptions of freedom vanished. This provocative finding

p-value Effect size 95% CI for effect size

<.001 ϕ = 0.58 [0.62, 2.23]

<.001 ϕ = 0.22 [0.20, 0.71] .11 ϕ = 0.10 [−0.05, 0.45]

.19 d = 0.45 [−0.21, 1.11]

<.001 d = 0.46 [0.21, 0.72]

NS O

s c

P

l l 1 i p s e t p t b o o b

c q b s s e s i l a r o

i w m t f s o t p

o

i t r c e a e r w A t

d b i T e d w s T w q A i t f i t p f w

a t s t a c a t n

P

w a a i a i p t

PERCEPTIO

uggests that perhaps decision makers tasked with evaluating ustody should be urged to adopt the suspect’s perspective.

aradigm Strengths and Limitations

At the outset, it is important to comment on three possible imitations to generalizability. The first limitation concerns the ength of our interviews and interrogations. Relative to Phase

participants, who reported no urgent need to escape, suspects n real interrogations are likely to experience a greater com- ulsion to leave—yet less freedom to do so. In the U.S., most uspects are questioned for 30 min to an hour (Feld, 2013; Kassin t al., 2007; Leo, 1996; witness interviews are not included in hese samples). Bound by ethics guidelines, we could not expose articipants to undue stress of this duration. One might expect hat longer sessions would increase decision-making conflict by oth increasing the motivation to terminate and exacerbating ne’s subjective sense of custody. For now, however, the effect f interrogation time remains an empirical question that can only e tested within a far less stressful paradigm.

The second limitation concerns the absence of a guilty ondition. We created a paradigm in which participants were uestioned not about a mock crime role-played as part of a study, ut rather about an apparent real theft bearing real potential con- equences. We chose this approach to mimic as best we could the takes of a criminal investigation conducted not by a psychology xperimenter but by a person of authority. For obvious ethics rea- ons, we did not seek to induce randomly assigned participants nto actually stealing a wallet, a real crime—the guilty equiva- ent of the condition we had devised. Using other approaches, dditional research is therefore needed to examine whether our esults for innocent persons speak to subjective custody in actual ffenders.

A third limitation concerns our participants. In both stud- es, participants were college students at an urban university ho were, therefore, more educated than the average citizen. A ajority were women and were Hispanic. Moreover, because of

he inherently stressful nature of the task, all were pre-screened or trait anxiety and a history of anxiety disorders. Within our ample, participants were randomly assigned to the interview r interrogation condition. Still, additional research is needed o replicate our overall pattern of results within the broader

opulation.1

Despite its shortcomings, this paradigm offers a high level f experimental realism. Although Phase 1 sessions were con-

1 Demographic differences were found between actors and observers, rais- ng the question of whether these differences influenced the results. To assess his possibility, we ran a series of ANOVAs with observer ethnicity and inter- ogation/interview condition as independent variables, to test for effects on the ontinuous and dichotomous free-to-leave measures for both studies. We tested thnicity coded as a 5-level factor (Asian, Black, Hispanic, White, and “other”) nd also as Hispanic versus non-Hispanic. We found no evidence that observer thnicity influenced either measure in either study. However, due to the wide ange of cell sizes for each ethnic group, the homogeneity of variance assumption as severely violated, so these results should be interpreted with some caution.

more demographically balanced purposive sample would be required to assess his issue more rigorously.

b C t s m t a t i i

s d t r

F CUSTODY 395

ensed and brief in duration, a theft was staged that was elievable, engaging, and at times quite stressful—and seen by nnocent participants as a real, potentially high-stakes situation. his aspect of the study is also worthy of comment. Even after xcluding from participation those prone to anxiety, the tension uring these sessions was often palpable (e.g., one participant anted to phone her father, another called the guard a racist,

everal emptied their bags in an effort to prove their innocence). his part of the process was contained in time, however, and care as taken during debriefing. Participants in both studies were ueried after their sessions but before they were fully debriefed. t that point, 58.33% spontaneously described the experience

n positive terms; 28.13% were neutral; 12.5% were negative. In he latter category, one participant confided that it was “horrible or me emotionally, but realistic, because anyone could be put n a situation like that at any time.” In this regard, most saw he experience as interesting or enlightening, believable, well lanned, and realistic. Participants would likely have reported ar more negative affect, even after debriefing, if the sessions ere much longer. It is particularly noteworthy that the interview condition fully

dhered to the courts’ objective criteria for noncustodial ques- ioning. For that reason it is striking that participants in these essions did not feel freer to leave than those in the interroga- ion condition. Nor did they feel free to leave when explicitly dvised as such. Given that interviews were conducted by a ivilian guard on a college campus and not by a detective at

police station, our results may well underestimate the extent o which real suspects feel free to terminate situations deemed oncustodial by law.

ractical Implications and Future Directions

In Study 1, a clear divergence in perceptions was obtained herein observers believed actors felt free to leave when the

ctors themselves did not. Study 2 showed that even when actors re explicitly advised that they could leave, which significantly ncreased the belief that they were in fact free to leave, that dvisement did not significantly increase their subjective feel- ngs of freedom. In the absence of evidence that advisement recludes feelings of custody in actors, a finding worthy of addi- ional research, police and the courts should be cautious of their elief in the efficacy of this practice (cf. Weisselberg, 2008). onsistent with the classic actor–observer effect in attribution,

his divergence raises questions about using a reasonable-person tandard in evaluating custody. With judges making the deter- ination from an observer’s perspective, our results suggest

he troubling possibility that subjective freedom will be over- ttributed to suspects, thereby curtailing their essential rights o Miranda warnings nationwide and the video recording of nterrogations in states that mandate this practice for custodial nterrogations.

Consistent with past research in non-legal domains, results

uggest a simple but potentially potent way to reduce this ivergence: instruct judges who rule on custody to imagine hemselves in the suspect’s position. In our studies, this personal eframing of the question closed the divergence gap by bringing

NS O

o s o s t A u T o i 2 H

b F c a t w h f p p p o n d p p

d b t h u p c o i a h o o

F y d s f

A a l

A

B

B

B

C D

D

F

G

G

G

H

H

H I

I

J

J

PERCEPTIO

bservers in line with the actors themselves. The resulting ynchrony in perceptions may stem from an alteration in the bserver’s visual and cognitive orientation, leading them to ee the impinging situation from the suspect’s perspective, hereby eliminating differences in attribution (Storms, 1973). lternatively, it may emerge from an increase in empathic nderstanding (Davis, Conklin, Smith, & Luce, 1996; Regan & otten, 1975), which can increase self-related cognitions about thers (Davis et al., 2004) and facilitate an understanding of ntergroup discrimination (Todd, Bodenhausen, & Galinsky, 012) and another person’s opposing attitudes (Tuller, Bryan, eyman, & Christenfeld, 2015). Given the implications for law, policy, and practice, we

elieve that future research should proceed in two directions. irst, although Phase 1 actors (and perspective-taking observers) onsistently reported feeling captive during interviews even after dvisement, the reason for this effect is less clear. Perhaps what hey lacked was the motivation of urgency because the session as brief, giving precedence to alternative motivations (e.g., to elp, to receive credit, to not appear evasive). With all actors actually innocent and perceived as such by observers, it is also ossible that because of the phenomenology of innocence, some articipants failed to grasp the potentially perilous nature of their redicament (Guyll et al., 2013; Kassin & Norwick, 2004; for an verview, see Kassin, 2005). In addition to unpacking the mecha- ism underlying this effect in self-report, future research should evise behavioral measures of custody by examining whether articipants actually attempt to terminate, or not, after lengthier eriods of time and when they are highly motivated to do so.

The second direction is to examine perceptions of custody in ifferent Phase 2 populations. The striking divergence obtained etween actors and observers is surprising when one considers hat both populations consisted of adult laypersons. In reality, owever, the custody inquiry is decided by a trial judge—a pop- lation of observers far removed from the “average” reasonable erson and yet to be tested. Similarly, future research should ast as observers other life-relevant populations such as police fficers (who make real-time custody determinations in decid- ng whether to Mirandize a suspect or record the interrogation) nd juveniles (a population prone to compliance, poor compre- ension of Miranda, and a high risk of false confession). It is ur hope that this research serves as a starting point in the study f custody.

Conflict of Interest Statement

The authors declare no conflict of interest.

Author Contributions

All authors contributed to the study conception and design. .A. and T.L. conducted data collection and performed the anal-

ses. All authors contributed to interpretation of analyses. F.A. rafted the manuscript, and S.K. and T.L. provided critical revi- ions. All authors approved the final version of the manuscript or submission.

F CUSTODY 396

Acknowledgments

The authors would like to thank Rebecca Cheiffetz, Aria mrom, Johanna Hellgren, Natalie Simonelli, Hyun-Joo Lee,

nd Sallyann Anderson for assisting significantly with data col- ection.

References

belson, R. P. (1981). Psychological status of the script concept. Amer- ican Psychologist, 36, 715–729.

eck, A. T., Ward, C. H., Mendelson, M., Mock, J., & Erbaugh, J. (1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561–571.

eirne, P. (1987). Adolphe Quetelet and the origins of positivist crim- inology. American Journal of Sociology, 92(5), 1140–1169.

uhrmester, M., Kwang, T., & Gosling, S. D. (2011). Amazon’s Mechanical Turk a new source of inexpensive, yet high-quality, data? Perspectives on Psychological Science, 6(1), 3–5.

alifornia v. Beheler, 463 US 1121 (Supreme Court July 6, 1983). avis, M. H., Conklin, L., Smith, A., & Luce, C. (1996). Effect of

perspective taking on the cognitive representation of persons: A merging of self and other. Journal of Personality and Social Psy- chology, 70, 713–726.

avis, M. H., Soderlund, T., Cole, J., Gadol, E., Kute, M., Myers, M., & Weihing, J. (2004). Cognitions associated with attempts to empathize: How do we imagine the perspective of another? Personality and Social Psychology Bulletin, 30, 1625–1635. http://dx.doi.org/10.1177/0146167204271183

eld, B. C. (2013). Real interrogation: What actually happens when cops question kids. Law & Society Review, 47(1), 1–36. http://dx.doi.org/10.1111/lasr.12000

ilbert, D. T., & Malone, P. S. (1995). The corre- spondence bias. Psychological Bulletin, 117, 21–38. http://dx.doi.org/10.1037/0033-2909.117.1.21

rice, P. (1989). Studies in the way of words. Cambridge, MA: Harvard University Press.

uyll, M., Madon, S., Yang, Y., Lannin, D. G., Scherr, K., & Greathouse, S. (2013). Innocence and resisting confession dur- ing interrogation: Effects on physiologic activity. Law and Human Behavior, 37, 366–375. http://dx.doi.org/10.1037/lhb0000044

arvey, J. H., Harris, B., & Barnes, R. D. (1975). Actor–observer differences in the perceptions of responsibility and freedom. Journal of Personality and Social Psychology, 32, 22–28. http://dx.doi.org/10.1037/h0076852

eider, F. (1958). The psychology of interpersonal relations. New York: Wiley.

owes v. Fields, 132 S. Ct. 1181 (Supreme Court 2012). nbau, F. E., & Reid, J. E. (1962). Criminal interrogation and confes-

sions (1st ed.). Baltimore: Williams and Wilkins. nbau, F. E., Reid, J. E., Buckley, J. P., & Jayne, B. C. (2013). Criminal

interrogation and confessions (5th ed.). Burlington, MA: Jones & Bartlett Learning.

ones, E. E. (1990). . Interpersonal perception (Vol. xiv) New York, NY, US: W H Freeman/Times Books/Henry Holt & Co.

ones, E. E., & Nisbett, R. E. (1972). The actor and the observer:

Divergent perspectives of the causes of behavior. In E. E. Jones, D. E. Kanouse, H. H. Kelley, R. E. Nisbett, S. Valins, & B. Weiner (Eds.), Attribution: Perceiving the causes of behavior. Morristown, NJ: General Learning Press.

NS O

K

K

K

K

L

M O O R

R

S

S S

S

T

T

W

W

Y

Received 26 July 2017; received in revised form 3 March 2018;

accepted 3 March 2018

PERCEPTIO

assin, S. M. (2005). On the psychology of confessions: Does inno- cence put innocents at risk? American Psychologist, 60, 215–228. http://dx.doi.org/10.1037/0003-066X.60.3.215

assin, S. M., Drizin, S. A., Grisso, T., Gudjonsson, G. H., Leo, R. A., & Redlich, A. D. (2010). Police-induced confessions: Risk fac- tors and recommendations. Law and Human Behavior, 34, 3–38. http://dx.doi.org/10.1007/s10979-009-9188-6

assin, S. M., Leo, R. A., Meissner, C. A., Richman, K. D., Colwell, L. H., Leach, A.-M., et al. (2007). Police interviewing and interroga- tion: A Self-report survey of police practices and beliefs. Law and Human Behavior, 31, 381.

assin, S. M., & Norwick, R. J. (2004). Why people waive their “Miranda” rights: The power of innocence. Law and Human Behav- ior, 28, 211–221.

eo, R. A. (1996). Miranda’s revenge: Police interrogation as a confidence game. Law & Society Review, 30(2), 259–288. http://dx.doi.org/10.2307/3053960

iranda v. Arizona, 384 US 436 (1966). regon v. Mathiason, 429 US 492 (1977). rozco v. Texas, 394 US 324 (1969). egan, D. T., & Totten, J. (1975). Empathy and attribution: Turning

observers into actors. Journal of Personality and Social Psychology, 32, 850–856.

oss, L. (1977). The intuitive psychologist and his shortcomings.

Advances in Experimental Social Psychology, 10, 173–220.

malarz, L., Scherr, K. C., & Kassin, S. M. (2016). Miranda at 50. Current Directions in Psychological Science, 25, 455–460. http://dx.doi.org/10.1177/0963721416665097

F CUSTODY 397

tansbury v. California, 511 US 318 (1994). torms, M. D. (1973). Videotape and the attribution pro-

cess: Reversing actors’ and observers’ points of view. Jour- nal of Personality and Social Psychology, 27, 165–175. http://dx.doi.org/10.1037/h0034782

ullivan, T. P. (2016). States which record statewide. Unpublished doc- ument (prepared July 11, 2016).

odd, A. R., Bodenhausen, G. V., & Galinsky, A. (2012). Perspective taking combats the denial of intergroup discrimination. Journal of Experimental Social Psychology, 48, 738–745.

uller, H. M., Bryan, C. J., Heyman, G. D., & Christenfeld, N. J. S. (2015). Seeing the other side: Perspective taking and the moderation of extremity. Journal of Experimental Social Psychology, 59, 18–23. http://dx.doi.org/10.1016/j.jesp.2015.02.003

atson, D. (1982). The actor and the observer: How are their percep- tions of causality divergent? Psychological Bulletin, 92, 682–700.

eisselberg, C. D. (2008). Mourning Miranda. California Law Review, 96(6), 1521–1601.

arborough v. Alvarado, 541 US 652 (Supreme Court 2004).

Available online 10 April 2018