Remaining silent during interrogation

Abstract

In many Western jurisdictions, criminal suspects undergoing police interrogations have the right to remain silent. In this experiment, we examined the effects of remaining silent during police questioning on laypersons’ perceptions of a suspect. Participants (N = 126) read one of three mock-interview transcripts (i.e. admission, denial or silence) and indicated the extent to which they agreed or disagreed that a male suspect in a missing person case was guilty, cooperative, trustworthy and rational. Participants expressed stronger agreement that the suspect was guilty when he admitted guilt than when he denied involvement or remained silent. When the suspect remained silent, participants viewed the suspect as less cooperative than when the suspect denied or admitted guilt and as less rational than when the suspect denied committing the crime. Our findings provide some support for the notion that remaining silent during police questioning may be viewed unfavourably by external observers.

Most Western countries have interrogation-related legal protections for criminal suspects undergoing custodial police questioning. These protections typically include both a right to silence and a right to legal assistance and are designed to address the inherent power imbalance between the investigator and the suspect (Kassin et al., 2020; Patry et al., 2014). To date, much of the literature on interrogation-related rights has focused on the degree to which suspects understand, and whether they choose to waive, their rights to silence and counsel (see Eastwood et al., 2014). Research in this area generally indicates that such rights are widely misunderstood and routinely waived during questioning, suggesting that they do not provide the intended protections for suspects (Eastwood et al., 2015; Kassin et al., 2020; Smalarz et al., 2016).

What remains relatively underexamined, however, is whether and to what extent a suspect’s decision to exercise their legal rights during police interrogation influences the perceptions and judgements of third-party observers (e.g. potential jurors, judges). In many jurisdictions, such as Canada, suspects have a legal right to silence such that no adverse inferences about the suspect can be made if they choose to invoke their right to remain silent (e.g. R v. Hebert, 1990). The extent to which people draw such inferences when reviewing suspect interviews is therefore an important empirical question. In the current experiment, we examined whether and to what extent a suspect’s silence influenced participants’ perceptions of the suspect.

Perceptions of the right to silence

Silence plays an important role in interpersonal communication generally and is of particular significance within police interrogations (see Ainsworth, 2012). The Supreme Court of the United States noted more than 50 years ago that suspects undergoing interrogation may assume that ‘silence in the face of accusation is itself damning’ (Miranda v. Arizona, 1966, p. 468). More recent empirical data suggest that this is a persistent concern. For example, Rogers and colleagues (2010) found that about one third of a sample of pretrial defendants and college student respondents believed that remaining silent could be used as incriminating evidence in a future trial; nearly 10% of pretrial defendants believed that the longer one remains silent, the more charges police will add. More recently, Mindthoff and colleagues (2018) found that jury-eligible U.S. participants – both community members and students – believed that suspects would waive their Miranda rights due to a concern about appearing guilty to third parties (e.g. police, jurors). Participants also agreed that suspects would think that if they talk, they could convince the police that they are innocent.

Do these beliefs match the perceptions of actual third parties? There is some evidence that third-party observers do indeed perceive silence as incriminating. In one study, mock jurors read a transcript of a murder trial in which a male defendant either (a) invoked the Fifth Amendment (i.e. the right to remain silent to avoid self-incrimination) in response to crime-specific questioning, (b) invoked the Fifth Amendment by not taking the witness stand, or (c) took the stand and answered questions (Shaffer & Case, 1982). Participants viewed defendants who invoked the Fifth Amendment – either by declining to answer specific questions on the stand or by declining to take the stand entirely – as more likely to be guilty and more deserving of conviction than defendants who did not.

In two unpublished experiments by Sukumar and Kassin (2017, as cited by Kassin et al., 2020), participants read a transcript of an interrogation of a male burglary suspect. In all conditions, the suspect was read his Miranda rights and initially denied involvement in the crime. In response to further interrogation, either the suspect agreed to speak and continued to deny the allegations or he invoked his right to silence, responding to all remaining questions with ‘no comment’. Participants were more likely to view the suspect as guilty when he invoked his right to silence than when he continued to respond with denials. These authors replicated this finding in a second experiment in which this effect persisted even after providing participants with additional instructions (i.e. reminding them of a suspect’s right to silence and cautioning them against drawing an adverse inference).

In a replication and extension of Sukumar and Kassin’s (2017, as cited by Kassin et al., 2020) research, Geven and colleagues (2020) found that when a suspect remained silent, they were rated as less honest and less cooperative than when the suspect denied involvement. However, they did not find a significant difference in guilt ratings between when a suspect remained silent and when he did not (p = .068, Cohen’s f = 0.16). In another series of experiments, DeCelles and colleagues (2021) examined the effects of a suspect’s anger as compared to alternative responses, such as remaining silent, on their perceived guilt. Participants rated those who remained silent as guiltier than those who calmly denied the accusation and at least as guilty as those who responded angrily. Similarly, in situations other than the evaluation of criminal suspects, such as when considering people’s dating and employment preferences, those who withhold information are judged less favourably than those who reveal information (John et al., 2016).

Based on the evidence reviewed above, a suspect’s silence during a police interview may convey information via pragmatic implication. That is, a suspect’s silence may lead external observers to draw a pragmatic inference (i.e. that the suspect is guilty) but not one that is logically necessary (see Harris & Monaco, 1978). People may ‘read between the lines’ to draw a connection between a suspect’s decision to remain silent and the likelihood of the suspect’s guilt. Several studies have examined pragmatic implication within police interviews, generally focusing on the ways in which interviewers may communicate information indirectly or implicitly, such as by implying that the suspect will receive more lenient sentencing if they cooperate (e.g. Davis & Leo, 2012; Kassin & McNall, 1991; Luke & Alceste, 2020). Although the existing research has focused on the role of interviewer behaviour, questions remain as to what a suspect’s behaviour might also pragmatically imply to third-party observers. Here, we consider the suspect’s decision to remain silent – a course of action for which there is explicit legal protection, which is often advised by attorneys, and which is reportedly undertaken in practice by both guilty and innocent suspects (e.g. May et al., 2022).

The current study

There is some evidence that external observers may view a suspect’s silence as an indication of guilt (e.g. DeCelles et al., 2021; Sukumar & Kassin, 2017, as cited by Kassin, 2020), although, at present, the empirical literature remains underdeveloped. Elsewhere, researchers have considered the role of pragmatic implication in police interrogations, generally viewing interrogators as ‘senders’ of implicit messages (e.g. implied leniency; Kassin & McNall, 1991). Whether the suspect’s behaviour, such as their decision to remain silent, can also convey implicit information (e.g. regarding the likelihood that they committed the crime) has received considerably less research attention in the interrogation literature. Previous research has compared participants’ perceptions when a suspect remains silent versus when a suspect denies the allegation (DeCelles et al., 2021; Geven et al., 2020; Sukumar & Kassin, 2017, as cited by Kassin et al., 2020). Because denial is not the only alternative to remaining silent, we chose to include an additional comparison condition in which the suspect provides an admission of guilt.

In the current experiment we compared participants’ perceptions of a suspect who, during a police interview, either remains silent, denies the allegation or admits wrongdoing. Based on the available findings, we hypothesized that participants would rate the suspect as more guilty and less rational when the suspect remained silent or provided an admission than when the suspect denied the allegation. We also hypothesized that participants would rate the suspect as less cooperative and less trustworthy when they remained silent or denied the allegation than when they provided an admission.

Participants

A total of 136 responses were recorded in the Qualtrics survey. We excluded those who failed to complete the study (n = 6) and those who failed the attention check (n = 4). The participants included in the final sample consisted of 126 undergraduate students from Ontario Tech University (78 females, 46 males, and 2 other). The mean age of the sample was 21.98 years (SD = 6.08). The sample represented a variety of ethnicities: most participants self-identified as White or Caucasian (n = 34), South Asian (n = 23), Asian (n = 14), Indian (n = 7), Chinese (n = 5), Filipino (n = 5), Pakistani (n = 4), Arab (n = 2), Bengali (n = 2) or Caribbean (n = 2). Of the participants included in the final sample, 14% were in their first year of study (n = 18), 30% in their second year (n = 38), 39% in their third year (n = 49), 16% in their fourth year (n = 20) and 1% in their fifth year or more (n = 1).

Materials and design

Research materials and data are available on the Open Science Framework (https://osf.io/zxpqt/). We used a single-factor between-participants experimental design. The independent variable consisted of three levels: (a) silence (i.e. the suspect invokes his right to remain silent during the interrogation); (b) admission (i.e. the suspect waives his right to silence and admits wrongdoing); (c) denial (i.e. the suspect waives his right to silence and denies wrongdoing).

The study materials consisted of an online survey (hosted at www.qualtrics.com). The first page of the survey consisted of an informed consent form that outlined the study’s details. On the following page, participants provided demographic information. Next, participants received the following instructions:

In the next section you will be provided with information about a hypothetical criminal scenario. Later in the study you will be asked a series of questions about this scenario, so please pay close attention to the material that follows.

Participants then received the following overview of the case on the fourth page:

On Friday June 21st, local and campus authorities were made aware that a female exchange student, Sarah Yang, who attended the local university had been missing for several hours. Upon receiving this information, local and campus authorities went door-to-door in the area she was believed to have last been seen, to ask people if they had information which may help locate her. During this process, authorities talked to Christopher Brant, a recent graduate of the university. Christopher stated that he had been at home all day playing video games on his computer and had not seen anything. Nothing came of the door-to-door questioning and the missing female was not located.

Three days later, during their investigation, the authorities came across CCTV footage in which Yang was shown walking down one of the campus streets. Following slowly behind her, was a vehicle matching Brant’s. Although the vehicle was slowly getting closer to Yang, both her and the vehicle exited the frame before any contact was made. This is believed to be the last footage of Yang prior to her going missing. In light of this information, a digital analysis was conducted on Brant’s computer. It was determined that he was in fact not active on his computer during the time Yang was believed to have gone missing.

Due to this information, Brant was asked to come down to the police station for formal questioning.

Upon arrival to the police station Brant was brought to an interview room and advised by Detective Gregory Lopez of the following rights:

• The right to remain silent

• The right to be told why you have been arrested or detained.

• The right to hire and instruct a lawyer.

• The right to know about the availability of duty counsel and legal aid.

• The right to speak with a lawyer, in private, as soon as possible.

After receiving this information, Brant signed a document indicating that he understood his rights.

The following page will show the entire interview between Detective Lopez and Christopher Brant. You will be asked about the contents of this interview, so please pay close attention.

Next, participants received a mock-interview transcript in which the suspect either: (a) provided an admission of guilt (admission; n = 42), (b) denied involvement in the crime (denial; n = 41), or (c) remained silent (silence; n = 43). In the admission condition, the suspect’s final utterance in the transcript was ‘I didn’t mean to kill her, I just wanted her to calm down’. In the denial condition, the suspect’s final utterance was ‘No. I have never met that girl in my life and I was home all day Friday’. In the silence condition, the suspect did not make an utterance at any point during the interaction. Only the suspect’s responses varied across conditions. The interviewers’ utterances were held constant.¹ Complete transcripts are available on the Open Science Framework (https://osf.io/zxpqt/).

After reading the mock-interview transcript, participants indicated the extent to which they agreed or disagreed on 7-point scales (1 = strongly disagree, 2 = disagree, 3 = somewhat disagree, 4 = neither agree nor disagree, 5 = somewhat agree, 6 = agree, 7 = strongly agree) with four statements about the suspect: ‘The suspect is guilty’; ‘The suspect is trustworthy’; ‘The suspect is cooperative’; ‘The suspect behaved rationally during the interview’. Participants also provided reasons for their ratings (open-ended) and completed a multiple-choice attention check (‘What did Christopher Brant (the suspect) initially tell the police he was doing on Friday, June 21^st?’). Last, participants received a debriefing form that included the researchers’ contact information.

Procedure

We received approval from the Research Ethics Board at Ontario Tech University prior to data collection. We recruited participants through the university’s online recruitment system. Interested and eligible participants signed up to participate and accessed the study link. After providing informed consent, participants worked through the Qualtrics survey outlined above. We designed the survey to randomly assign participants to an experimental condition. Upon completion, participants received debriefing information. The median study completion time was approximately 9 minutes. All participants received course credit as compensation.

Results

We conducted four between-subjects analyses of variance (ANOVAs) to examine differences on each of the dependent measures (i.e. guilt, cooperativeness, rationality, trustworthiness) as a function of suspect behaviour (i.e. admission vs. denial vs. silence). As the unstandardized scale ratings were not intrinsically meaningful, we chose to perform group comparisons using standardized effect sizes (i.e. Cohen’s d) based on independent-sample t-tests. Given the lack of direct research in this area, we also choose to interpret d values in accordance with Cohen (1988). Descriptive statistics are presented in Figure 1.

Figure 1.

Descriptive statistics as a function of experimental condition. Note: values are mean ratings on 7-point scales (from 1 = strongly disagree to 7 = strongly agree). Error bars are 95% confidence intervals.

There was a significant effect of suspect behaviour on ratings of guilt, F(2, 123) = 33.43, p < .001, partial η² = .35. Participants agreed more strongly that the suspect was guilty when the suspect provided an admission of guilt (M = 6.36, SD = 0.85) than when he offered denials (M = 4.59, SD = 1.20, d = 1.70, 95% CI [1.20, 2.20], where CI = confidence interval) or remained silent (M = 4.79, SD = 1.17, d = 1.04, 95% CI [1.04, 2.01]). These were large effects. The comparison of guilt ratings between silence and denial conditions revealed only a small effect, d = 0.17, 95% CI [–0.26, 0.60].

There was a significant effect of suspect behaviour on ratings of cooperativeness, F(2, 123) = 89.88, p < .001, partial η² = .59. Participants were less likely to agree that the suspect was cooperative when the suspect remained silent (M = 2.07, SD = 1.33) than when he provided an admission (M = 5.21, SD = 1.14, d = 2.53, 95% CI [1.96, 3.10]) or denial (M = 5.12, SD = 1.21, d = 2.40, 95% CI [1.83, 2.95]). These were large effects. The comparison of cooperativeness ratings between admission and denial conditions revealed only a small effect, d = 0.08, 95% CI [–0.35, 0.51].

There was a significant effect of suspect behaviour on ratings of rationality, F(2, 123) = 7.29, p = .001, partial η² = .11. Participants expressed stronger agreement that the suspect behaved rationally when the suspect offered denials (M = 5.15, SD = 1.01) than when he remained silent (M = 4.02, SD = 1.49, d = 0.88, 95% CI [0.43, 1.32]). This was a large effect. The comparisons of rationality ratings between admission and denial conditions, d = 0.38, 95% CI [–0.06, 0.81], and between admission and silence conditions, d = 0.43, 95% CI [0.00, 0.86], revealed only small to moderate effects.

There was no significant effect of suspect behaviour on ratings of trustworthiness, F(2, 123) = 1.58, p = .21, partial η² = .03.

Discussion

Some evidence suggests that remaining silent in the face of alleged wrongdoing may be viewed as an indication of guilt (e.g. DeCelles et al., 2021). We examined this possibility in the current experiment by comparing participants’ perceptions of a suspect who either remained silent, denied involvement in the crime or provided an admission of guilt. Our first hypothesis was partially supported, as participants agreed most strongly that the suspect was guilty when he admitted guilt compared to when he denied committing the crime. Contrary to our hypothesis, however, there was minimal difference in perceived guilt between silence and denial conditions. Participants also viewed the suspect as less rational when remaining silent than when denying committing the crime. In line with our second hypothesis, participants viewed the suspect as less cooperative when he remained silent than when he made an admission. There was very little difference in ratings of cooperativeness, however, between suspects that denied the allegation and those that provided an admission.

These findings paint an interesting picture as it relates to the perception of suspects who choose to remain silent during interrogation. Contrary to our expectations, remaining silent did not appear to lead participants to make a pragmatic inference that the suspect was more likely to be guilty than when the suspect denied involvement in the crime. It did, however, lead to lower ratings of perceived cooperativeness and rationality. While it is reassuring to find that remaining silent does not increase perceptions of guilt to the same extent as directly admitting guilt, our results strongly suggest that a suspect who remains silent is likely to be perceived as uncooperative and less rational than a suspect who chooses to actively deny committing the crime. It appears that rather than viewing the strategy of remaining silent as a wise one to avoid incrimination (see Geven et al., 2020), participants viewed the approach in a negative light. Although speculative, participants may have questioned why the suspect did not provide denials or counterarguments to the provided evidence when given the opportunity during the interview. Given that the scenario used in the current study was relatively early in the investigative process (i.e. the initial part of the initial suspect interview), it is unclear how these negative perceptions may have impacted perceptions of the suspect as the investigative and judicial process unfolded. Overall, these findings suggest that remaining silent may lead to negative perceptions of the suspect and ultimately not provide the level of protection against incrimination that it was designed to provide.

Previous (unpublished) findings reported by Sukumar and Kassin (2017; as cited in Kassin et al., 2020) seem to show that remaining silent (i.e. responding with ‘no comment’) as compared to denial is associated with increased perceived guilt. However, this effect failed to replicate in a follow-up study (Geven et al., 2020). In the current study too, there was little difference between the silence and denial conditions in perceived guilt (partially consistent with DeCelles et al., 2021). It is important to note that the silence condition in the present study consisted of consistent silence throughout the entire interview. By contrast, Sukumar and Kassin’s silence condition apparently consisted of silence preceded by denial, such that the suspect initially denied committing the crime but then chose to remain silent upon further interrogation. Therefore, their comparison was denial only versus denial + silence, as opposed to denial only versus silence only. This is a subtle but potentially consequential difference; it may well be that remaining silent after denying involvement is more damning than remaining silent from the start. However, more research is needed to clarify this distinction.

Legal jurisdictions differ in how silence is interpreted. Even among countries that formally recognize a suspect’s right to silence, there are conflicting standards regarding whether external observers are permitted to draw adverse inferences from a suspect’s silence. In some jurisdictions, such as the United Kingdom, external observers are legally permitted to draw such inferences, at least under some circumstances (Criminal Justice and Public Order Act, 1994). Elsewhere, jurors are explicitly prohibited from drawing adverse inferences from silence (e.g. Canada; R v. Hebert, 1990). Legal differences aside, at some point in the adjudication process, external observers must make an inference of guilt, based partly on the interview interaction. To the extent that legal decisions are based on the content of the suspect’s interview, the present results suggest that the suspect’s silence may be a factor in this decision. However, questions remain as to the specific circumstances in which silence may have more or less of an impact on legal decision-making (e.g. as a function of crime type, evidence strength, suspect characteristics, non-verbal behaviour).

The current experiment has several limitations. First, because the police interviewer’s utterances were identical across conditions, our silence condition pertains to jurisdictions where it is legally permissible for police to continue questioning a suspect after they have exercised their right to remain silent (e.g. Canada). Our study design may not generalize to other jurisdictions where police questioning must cease once the suspect invokes their right to silence (e.g. United States).² Second, although the effect of suspect behaviour on perceived trustworthiness was nonsignificant, some participants may have believed that the evidence described within the case scenario was sufficiently indicative of guilt, such that any statements to the contrary (i.e. denials) were assumed to be unintentionally contradictory or intentionally deceptive. This was noted in the open-ended responses of several participants in the denial condition (e.g. ‘All the evidence is in favour of Brant being guilty, and it is clear he is lying even when the evidence is against him’; ‘his story didn’t match evidence they had . . . ’; ‘It seems that there is a sufficient enough evidence to point out that the suspect is lying about his whereabouts on the day the girl went missing . . . ’). This remains speculative, however; a direct manipulation of evidence strength in future research would help to shed light on this issue. Third, our participants read a mock interview transcript rather than viewing a recording. This allowed for a clean comparison of the three levels of suspect behaviour but at some expense to ecological validity.

Conclusion

The right to silence during police interrogation is widely recognized in many countries (Kassin et al., 2020). Although remaining silent is legally permissible and sometimes explicitly encouraged by legal counsel, the current results suggest that remaining silent offers little recourse as an alternative to continually denying committing the crime. In line with Geven and colleagues (2020) and in contrast to Sukumar and Kassin (2017; as cited by Kassin et al., 2020), we found that there was very little difference (d = 0.17) in the perceived guilt of a suspect who remains silent versus one who denies committing the crime. However, remaining silent was viewed as a less rational form of behaviour than denial and considerably less cooperative than either denial or admission. Our findings suggest that suspects who exercise their right to remain silent may be perceived as uncooperative, thus raising new questions regarding the protective value of existing interrogation-related rights.

Funding Statement

The first author was supported, in part, by the Social Sciences and Humanities Research Council of Canada [Joseph-Armand Bombardier Canada Graduate Scholarship # 752-2019-1596].

Declaration of conflicts of interest

Mark D. Snow has declared no conflicts of interest

Quintan Crough has declared no conflicts of interest

Cassandre Dion Larivière has declared no conflicts of interest

Funmilola Ogunseye has declared no conflicts of interest

Joseph Eastwood has declared no conflicts of interest

Ethical approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the Ontario Tech University Research Ethics Board (File No. 16730) and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed consent

Informed consent was obtained from all individual participants included in the study