5
$\begingroup$

Context

I am planning an experiment where the subjects would need to determine the language of the stimulus after listening to it once, given two alternatives, L1 or L2. In total I have 500 trials separated in 5 blocks.

Problem

One of my supervisors insists that I have to give the subjects a feedback immediately after they made the choice ("because the experiment is boring"), and the other one says I should not give them any feedback at all ("they get payed for this, they have to suck it up").

I, for myself, think that maybe giving them scores after the end of a 5-minute block would be helpful to support their motivation and give a little more of motivation for them to concentrate on the following block.

Question

Does anyone here have experience with giving feedback to subjects during the experiment?

$\endgroup$
3
$\begingroup$

Short answer
Giving subjects feedback may affect their performance over time because of learning effects. The choice of giving feedback may be far less trivial than subjects getting bored yes or no. And quite frankly, subjects will get bored after a 100 trials, let alone 500, feedback or no feedback...

Background
You have to be aware that subjects can show learning effects in speech tests (De Jong et al, in press). I am unaware of the specific task you give them, but perhaps below answer will help you make a rational decision on this.

  • Procedural learning affects the way people perform because they may get better in the task, simply because they have to learn how the task works ('Oh wait, the trial starts already!' 'Oh wait, where was that answer button again?' 'Oh wait, what was the green button for again?'). If subjects show procedural learning effects, they will get better over time. This is oftentimes regardless of whether you give any feedback; subjects simply need to 'get the hang of it'.
  • Conceptual learning will affect the performance too. This type of learning is strongly affected by whether you give them feedback. For example, if your subjects are unfamiliar with the languages and you are not providing them feedback then they will never learn them. On the other hand, when for example they hear words from a closed set and each word is repeatedly presented, and the experimenter provides the correct answer ('This trial you heard senorita and that is Spanish and means female' or whatever) and the subjects repeatedly hears this word then for sure at a certain point they will learn from feedback that it is Spanish. However, without any feedback they will never learn it.

Even feedback after 5 minutes may affect their performance ('Oh I'm doing great, I can probably relax a bit'). In my opinion, subjects should never be given feedback, not until, perhaps, after the entire experiment is over.

Reference
- De Jong et al., Ear & Hearing, in press

$\endgroup$
  • $\begingroup$ thank you, this is very helpful! I think I will make a test trial for them to compensate the procedural learning effect to some degree during the experiment. To make a bit more clear the idea: I am presenting them with pseudo-speech stimuli that are created out of 2 languages. One is their native (to which I expect them to show a high sensitivity) and the other is not familiar. $\endgroup$ – irishpippin Nov 16 '17 at 14:10
  • $\begingroup$ @Scio - thanks for the comment. A few practice trials are highly recommended for sure. $\endgroup$ – AliceD Nov 16 '17 at 14:17
2
$\begingroup$

Let me give the opposite advice. Note that I don't know the language field. But in perception people distinguish between so-called type A and type B experiments (or type 1, type 2, the classification is actually quite confusing as all authors have a different one). In type A experiments you know what the correct answer is (for example is this beep higher or lower than 1kHz). In type B experiments there is no way to know the correct answer as it depends on the subject (for example whether your subject is experiencing an aftereffect). So they answer slightly different questions, type A experiments ask "how good is my observer at answering my question?", type B experiments ask "how does my stimulus appear to the observer?". Therefore in perception people generally agree that if you do know the correct answer to a question, you should always provide it through feedback. It prevents observers from developing biases and allows you to truly estimate how good your observers could actually be in your task. But if what you want is precisely to study how language appear, then I guess it would be ok not to give feedback (but personally I would).

Edit: below a good reference, see chapter 2. The classification was originally introduced by Brindley, the same one from the infamous Las Vegas conference.

Kingdom, F. A. A., & Prins, N. (2010). Psychophysics: a practical introduction.

$\endgroup$
  • $\begingroup$ What do you think of that book btw? I bought it too for general reference, but I don't know if it's good or not. Why did you read it? Have you used it for classes? $\endgroup$ – AliceD Nov 23 '17 at 9:01
  • 2
    $\begingroup$ I believe the book is freely available (if not, it definitely is on some some russian website). My opinion is that Fred Kingdom and Nicolaas Prins are ones of the best psychophysicists, and their book is always 100% spot on. It's annoying at times that they only explain how their toolbox (palamede, which is awesome) work, instead on just explaining the general principle. But I would make the same argument for the Knoblauch & Maloney book (the R e $\endgroup$ – baca Nov 23 '17 at 10:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.