Confession Errors

Today’s Post suggests we rely too heavily on legal confessions: 

In one experiment, Kassin asked volunteers to perform a challenging task on a computer but warned them not to touch the "Alt" key or risk damaging a computer. Volunteers were told that the computer had been damaged and were asked whether they hit the banned key. In reality, the volunteer did nothing wrong. Most volunteers denied it, but as the initial task they were given was made difficult, they became less sure because they were distracted. When researchers had confederates lie about having seen the volunteers hit the Alt key, the number of people who confessed went up to 100 percent. Every stage of increased pressure led ever larger numbers of volunteers to believe they were really guilty.

GD Star Rating
loading...
Tagged as:
Trackback URL:
  • joe

    Robin,

    This situation seems to be very different than most legal confessions. In this setup, a person might honestly not know whether or not they committed a “crime.” Being given the information that the computer was fine before but is not now should be taken as evidence that maybe you did accidentally hit the key. And if in addition, a “witness” comes forward who you have no reason to believe is lying, why should you not believe him? In this situation, people who cannot honestly be confident in their innocence should rationally come to believe that they were guilty.

  • Doug S.

    My current view on legal confessions is that, if the confessor is not willing to stand by a confession in court, I will ignore it. I’ve just heard too many stories about confessions being wrong…

  • Nick Tarleton

    I suspect it would be MUCH, MUCH, MUCH easier to convince myself I did something (hit the Alt key) that I naturally do all the time, and had to make an effort to avoid, than to convince myself that I committed a serious crime.

  • Aaron

    Hitting the Alt key while doing a complicated project

    • Is something a person would plausibly do without conscious thought (e.g. accidentally or habitually),
    • Is easily forgotten (because it’s simple and commonplace),
    • Has relatively minor consequences (e.g. “damaging a computer” vs. a murdered colleague followed by 45 years in prison),
    • Is hard to deny (even if true) without looking irresponsible and antagonistic in most human societies.

    I gather that the researcher extracting confessions

    • Seemed like an ally with no obvious reason to lie,
    • Did not threaten any large punishment.

    Under these conditions, it’s hardly surprising that confessions could be extracted. It may be quite rational for the subjects to wrongly confess under those conditions.

  • josh

    Still interesting, though.

  • Constant

    I wouldn’t confess. This has been tested in life. Eventually somebody else “confessed” so that my interrogator would drop the subject.

  • Unnamed

    There are factors that promote false confessions in the lab study, but there are also factors that promote false confessions in real interrogations, including:

    – long interrogations
    – highly trained interrogators
    – interrogators’ goal is to extract a confession, not to figure out the truth (they assume that innocent people would never confess, or that they already know the guy is guilty)
    – an ambiguous relationship between the suspect and the interrogators (they often seem to be on your side, and make it seem like everything will be fine once you confess)
    – the interrogators claim to have independent evidence establishing the suspect’s guilt, and the suspect believes that they must be honest (when, in reality, they are free to lie)

    Often, the false confession starts with the interrogator presenting some (made-up) fact tying the suspect to the crime and asking “how could this be?” The suspect, assuming it must be true, starts trying to explain it. At that point, he is on his way to creating a plausible-sounding story, aided by the interrogator who will guide him along. Of course the jury will only see the finished product.

  • http://entitledtoanopinion.wordpress.com/ TGGP

    Unnamed, you’ve got consistently good posts. I recommend that you start an anonymous blog.

  • http://profile.typekey.com/sentience/ Eliezer Yudkowsky

    You can be “Unnamed” right here on Overcoming Bias if you like.

  • Douglas Knight

    Unnamed,
    What is the evidence for your claims? They sound plausible, but you imply more confidence than that.

    What is your claim about highly trained interrogators? In particular, what kind of training? training in extracting confessions? false confessions? true confessions, training for other police duties?

  • Unnamed

    I’m afraid that my evidence would not be admissible in court, since it counts as hearsay. Most of it comes from talking with people who do research in psychology & law, or watching them give presentations. Their sources include data on false confessions and training books for interrogators.

    Interrogators are trained to extract confessions, as described in the following bullet point.

    Glad to know my comments are appreciated, TGGP & Eliezer (and Robin in a previous post, I believe). At this point, I’d rather not put the time or commitment into blogging. I’m happy commenting occasionally, when I think I have something worth adding.

  • Douglas Knight

    I’m sorry if I came across as hostile. I wouldn’t say you overplayed your evidence and I wouldn’t call informal reports of systematic investigations “hearsay.” I would call my information hearsay with great danger of selection bias.