When investigating a crime, detectives often collect evidence by talking with people, such as an eyewitness that saw the crime, someone who can corroborate an alibi (well, hopefully), or a suspect who may have committed the crime. Of course, people’s statements to police aren’t always trustworthy. After all, if they committed the crime, they may not come right out and admit it. As a result, officers will sometimes use special interviewing techniques to attempt to distinguish truth-tellers from liars.
Psychological scientists have, for the last several years, been working to develop interviewing techniques designed to improve investigators’ judgments of who is lying and who is telling the truth1. But long before psychologists started developing such techniques, law enforcement practitioners developed their own tools, mostly based on experience and intuition, rather than scientific research. One such technique developed by practitioners is called a “bait question” – a tactic in which officers try to “bait” you into changing your story by suggesting the existence of incriminating evidence. As the reasoning goes, if you take the bait and change your story to fit the evidence, you are probably guilty2.
|If they change their story, they've taken the bait |
and fallen for the trap!
But does that mean they're actually guilty?
This is, in essence, an attempt to create a cue to deception. We talked last week about how naturally-occurring behavioral cues to deception aren’t strong enough to achieve high levels of accuracy, so if bait questions can create reliable and strong cues to deception (rather than simply looking for them), they would be a really useful tool. We want to make this very clear: We aren’t sure if bait questions work. But let’s take look at how a bait question is supposed to work…
The police will ask the person they’re interviewing if they can explain the existence hypothetical incriminating evidence. If the person is being truthful, they’re supposed respond that there’s no way the evidence can exist. If the person is lying about committing the crime, however, they’ll change their story to account for evidence that they assume exists. As the name suggests, the police use evidence as bait – if the person is guilty, they’ll supposedly take the bait and try to explain the evidence away.
For example, the police may ask “Is there any reason we would find your fingerprints of the gun used to murder the old woman?” An innocent person is expected to reply, “Nope, no reason at all.” However, if you’re guilty, you’re expected to respond differently: You need to explain why your fingerprints are on the gun, but why that doesn’t mean you committed the crime. So you’re expected to say something like, “Hm, well, I had a gun, but it was stolen a few weeks ago. So if whoever stole it used the gun in the murder, my fingerprints may still be on it.”
Let’s ask the crucial question: Do bait questions work at telling liars from truth tellers? To date, we know of only one study that has directly tested whether bait questions produce different responses from liars and truth-tellers – and that single study found that bait questions work3. The study’s results have never been replicated, so it’s hard to tell if this finding is reliable*. Additionally, other studies that have examined similar interviewing techniques that suggest the existence of evidence have found that such techniques can increase the risk that an innocent person falsely confesses4.
But regardless of their effectiveness, bait questions can have some problematic side effects. The two of us (your blog-hosts, Will and Timothy) have been working together on a research project for a few years, and in the course of our research, we’ve found bait questions can cause false memories for what evidence actually exists. Let’s set aside whether bait questions are a useful lie detection tool or not, and instead look at other results of using them. A huge problem with bait questions is the leap from “this person answered my question” to “this person committed the crime!” Officers are trained to ask bait questions about evidence that might not actually exist. Depending on what the bait evidence is, it’s perfectly reasonable that an innocent person may, perhaps in an attempt to help the police, answer the bait question and try to explain the evidence – a strategy which could backfire. How people process and respond to bait questions, and whether innocent and guilty people do this differently, has not been widely researched yet, however, so it’s difficult to say exactly under what circumstances people reject or explain the evidence.
But bait questions pose another risk beyond innocent people looking suspicious – they can actually distort peoples’ memory. One of the key qualities is that they ask about incriminating evidence that the police don’t actually have. Going back to our earlier example, the detectives may not have any fingerprints from the gun, let alone the suspect’s. But there’s a lot of psychology research that shows hearing something described, such as fingerprints on the gun, can lead people to think that it actually exists. For example, we’ve discussed previously (link) that if you listen to someone inaccurately describe something you saw, you’re likely to remember what they said rather than what you actually saw – a phenomenon called the misinformation effect5. In our research, we’ve found that bait questions can have this same effect – people who hear the bait question come to remember the hypothetical, non-existent evidence actually exists, despite being told otherwise6.
We discovered this by having participants read a set of “case facts” about a case we made up. The case facts describe the robbery of a liquor store, and all the evidence that was collected in the investigation – including some key pieces of evidence linking the suspect to the crime, such as his fingerprints on a weapon used during the robbery, the crowbar used to pry open the cash register found in his car, and an eyewitness placing him at the scene. Then, after a few minutes wait, participants watched a recorded interrogation that we filmed. In the interrogation, the police officer asked the suspect (both actors, and not a real officer and suspect) a bait question about some of the evidence they recovered. Some of the evidence was referred to in a neutral, general way, so that the participants’ memory wasn’t refreshed. So the officer would ask “Is there any reason we would find a metal tool [instead of crowbar] used to open the cash register in your car?”**
|How was the cashregister opened? Was it a screwdriver, or a crowbar?|
A bait question could make you make the wrong choice
For other pieces of evidence, they asked the question about similar, evidence that wasn’t mentioned in the report (and thus was never collected), such as “Is there any reason that we would find your skin cells [instead of fingerprints] on the weapon recovered from the scene?”
After watching the interrogation film, participants had their memory tested. We asked them about evidence recovered from the scene and which evidence was actually collected by the police. Compared to the neutrally-worded bait questions, which serve as a control measurement of memory, participants were more likely to misremember what evidence actually existed when they were asked bait questions. Remember the above example, where one bait question was neutral (“crowbar” and “metal tool”) and one was misleading (“fingerprints” and “skin cells”). In this situation, participants were pretty accurate about remembering a crowbar was used to open the cash register (they got it right about 80% of the time). But only about 60% of people accurately remembered fingerprints were found on the gun – whereas almost 40% of participants said skin cells were recovered. To date, we’ve tested over 1,000 people in different versions of this study, and we’ve reliably observed these effects.
|Memory Accuracy Means from one of our Bait Question studies.|
A score of 100% means perfect memory. As you can see,
memory for the Misled Items is worse - showing bait questions
are distorting memory for evidence.
What this means is that our participants were suffering false memories for what evidence was real. Instead of accurately remembering what was described in the report, they sometimes remembered the evidence asked about in the bait question – evidence that didn’t actually exist and wasn’t collected. In fact, participants generally remembered more evidence than was actually collected – participants, on average, recalled police collected more than 12 pieces of evidence incriminating the suspect, even though the case report only described 10. That means that in some cases, participants were thinking both pieces of evidence were collected – e.g., remembering the police collected both the fingerprints and skin cells from the gun.
That’s definitely bad news, and it gets worse. In follow up studies, we warned participants that some of the evidence in the interrogation didn’t actually exist – either before they watched the interrogation or after. For some participants, we pointed out which evidence didn’t exist, such as saying “The officer asks about evidence found on the gun. That evidence was not actually collected by the police.” We reasoned that in the real world, attorneys might use similar strategies to point out such misinformation in a trial. Unfortunately, these warnings had no effect – even when we told participants which evidence didn’t exist, they still showed the same rate of false memories for evidence.
There are a couple possibilities for why the warnings didn’t reduce false memory rates. First, it may be that once someone has determined the suspect is guilty, the specific evidence doesn’t matter as much and they don’t pay as close attention. Other research has shown that juror thinking can be bi-directional8 – that is, people evaluate whether someone is guilty based on evidence, but once they’ve decided someone is guilty, they then evaluate the quality of evidence based on whether it supports guilt. Here, it may be that exactly what form of evidence is found doesn’t matter – both fingerprints or DNA indicate guilt.
Another possibility is that trying to keep track of all the evidence and details from the case is a task that requires a lot of cognitive resources. Perhaps trying to remember all of the evidence and decide how it relates to the rest of the crime is so taxing that keeping in mind the warning while considering each piece of evidence is too much cognitive work to do successfully. Other research has shown that warnings similar to what we used normally work with similar memory tests7 – that is, people can often sort out real information from misleading information if they’re warned about it. But perhaps there’s something a bit unique about legal decision making, crime stories, or hypothetical evidence that makes this task more difficult.
We may not know exactly why our warnings didn’t work, but the overall finding that bait questions can form false memories for evidence that are extremely resistant to correction is very troubling. Of course, this was a psychology study, not a real life case. But these findings are fairly telling for what may happen if bait questions are used in the real world, and have important implications for actual trials.
First of all, our participants here were like jurors – they weren’t the person being interrogated, they were merely reviewing all of the evidence from the case. We don’t know how bait questions might affect the memory of the officer asking the questions or the suspect answering the questions (although we definitely want to find out!). But with many experts advocating for video recording of interrogations, it’s likely that what happens in the interrogation room won’t always stay in the interrogation room – both prosecutors and defense attorneys (for different reasons) may begin using recorded interrogations as evidence in trial.
A potential side effect of showing interrogations in trial is that the specific evidence mentioned in the interrogation questions may not be subject to the same legal rules on admissibility as real evidence. For example, polygraph results are usually not admissible in court – but if an officer asks a bait question about a failed polygraph, jurors may come to remember the failed test as actually existing. Similarly, officers are not allowed to fabricate and show suspects fake forensic reports (such as a DNA test). But if that same evidence is the topic of a bait question, jurors could misremember it actually exists – so it’s practically in evidence anyway.
|The suspect's fingerprints were at the crime!|
...he must be guilty, right?
But our findings are problematic because jurors are supposed to be reliable fact-finders in the legal system. Jurors consider all of the facts and theories, and are supposed to divine the truth about what happened. If a suspect is trying to help the police out and be cooperative to prove their innocence, they’ll look like they’re lying to the officer. If that case goes to trial, we would hope the jury would realize the bait question is hypothetical and not use the hypothetical evidence against the suspect. But our research suggests a different result: Jurors may actually believe the evidence does exist and come to remember it was collected by the police.
In our first set of studies, we did not look at whether these false memories for evidence result in participants thinking the suspect is more likely to be guilty. We are, however, looking at this in another set of studies currently on-going though (osf.io/z6u3e ; osf.io/4mn5h). So far, we’ve found that it isn’t as cut and dry as “more false memories = higher guilty verdicts.” But it seems that if a participant falsely remembers the hypothetical evidence in a bait question, they will see a weak case built on very little circumstantial evidence as a strong case built on solid incriminating evidence, or change their understanding of a crime to fit the evidence into a situation where the suspect is guilty.
One of the most interesting things about the problems of memory for bait questions it that they are all side-effects. That is, an officer could be using bait questions completely correctly, but still causing memory distortions. The take-away from this is two-fold. First, we should consider the variety of costs and benefits to any investigative technique. If it turns out that bait questions are effective in helping us find liars, but can cause jurors (and perhaps others) to misremember what evidence actually exists, are they worth using?
Second, this underlines the importance of an understanding of psychology applied to the law. That bait questions can affect memory is perhaps not immediately obvious to legal practitioners – however, to a memory expert, bait questions sound exactly like four decades worth of research that demonstrates asking questions can change what someone remembers. Thus, this reinforces the importance of science in evaluating how things work in applied situations.
There is still a lot we don’t know about bait questions – and even more we don’t know about how other similar tactics may affect memory and the legal process. However, our initial results suggest that further research is warranted and care should be taken when using this investigative strategy.
This post was written by Will Crozier / edited by Timothy Luke
* There are other, more technical, reasons for treating this study’s results with caution. When researchers make a lot of statistical comparisons, sometimes they find effects that look real but are actually just the result of random error – if you throw a hundred darts at a board, a few of them are bound to hit the bullseye. In the experiment, the bait question was one of 16 different special “behavior-provoking” interview questions tested. The bait question was the only one that appeared to work as intended (all the others didn’t work at all, or made innocent participants look guiltier than guilty participants). So the bait question may have just been a lucky throw, so to speak. Without studying the bait question a lot more, we won’t know for sure.
**Incidentally, writing research materials that asked this question a bunch of times had an odd effect on us: Any time someone asks "Is there any reason..." or "Is it possible..." in normal conversation, we can't help but laugh.
 Hartwig, M., Granhag, P. A., & Luke, T. J. (2014). Strategic use of evidence during investigative interviews: The state of the science. In D. C. Raskin, C. R. Honts, & J. C. Kircher (Eds.) Credibility assessment: Scientific research and applications (pp.1-36). Amsterdam, NL: Elsevier.
 Inbau, F. E., Reid, J. E., Buckley, J. P., & Jayne, B. C. (2013). Criminal interrogations and confessions (5th ed.). Burlington, Massachusetts: Jones & Bartlett Learning.
 Vrij, A., Mann, S., & Fisher, R. P. (2006). An empirical test of the behavioral analysis interview. Law and Human Behavior, 30, 329–345.
 Perillo, J. T., & Kassin, S. M. (2011). Inside interrogation: The lie, the bluff, and false confession. Law and Human Behavior, 35, 327–337.
 Loftus, E. F. (2005). Planting misinformation in the human mind: A 30-year investigation of the malleability of memory. Learning & Memory, 12(4), 361-366.
 Luke, T. J., Crozier, W. E., & Strange, D. (2017). Memory errors in police interviews: The bait question as a source of misinformation. Journal of Applied Research in Memory and Cognition, 6(3), 260-273.
 Blank, H., & Launay, C. (2014). How to protect eyewitness memory against the misinformation effect: A meta-analysis of post-warning studies. Journal of Applied Research in Memory and Cognition, 3(2), 77-88.
 Simon, D. (2004). A third view of the black box: Cognitive coherence in legal decision making. The University of Chicago Law Review, 511-586.