Question Regarding MLDS method from your Psychophysics book

[gsharma70]

First of all, thank you very much for taking the time to write this book as well as the PAL toolbox. It has been an extremely valuable resource for us in trying to design psychophysics based experiments.

I wanted to ask a few related questions in two parts.

Part 1:
We are designing a haptic based experiment using the MLDS approach described in your book. We are using (N=3, or triads) for each trial. However, in terms of user responses for each trial, we were slightly unsure about which among the ones below would be the correct question to ask the participant;
1. Is the first pair of stimuli more different than the second pair?
2. Which pair, 1st or 2nd is more different?

Although similar, they do have slightly different implications according to our view. For instance, if both pairs appear the same to the user, the answer to question 1 would be an unambiguous NO. However, for the second question, there is a 50% chance to say 1 or 2. Over the large number of trials that MLDS requires, we believe this would have implications on the resulting perceptual scale. So, we would love to hear your thoughts on the following;

Do you think our thinking is right in regards to (1) and (2) having different implications?
Regardless of our thinking, is there a correct question to ask while designing the experiment or does it not matter?

Part 2:
I noticed from the literature that MLDS has been applied in various vision based experiments that generally do not have to consider temporality as a factor. Eg: All four stimulus are presented in the screen at once and a participant responds based on that. However, in the context of haptic based experiments, each stimuli, even within a single trial have to be presented in a certain temporal order. One of the concerns we have is that, let's say by the time we present the second pair of stimuli and ask the question "which one is more different", participants might have somewhat forgotten about the first pair, thus introducing additional noise to the responses. With that said, I think large number of trials along with randomization helps mitigate this possible temporal bias. I would really love to hear your thoughts on this and/OR if you know of a resource that tackles this problem, it would be great if you could point me to it.

Thank you very much!

Best,
Gyanendra

[Frederick Kingdom]

Part 1:
We are designing a haptic based experiment using the MLDS approach described in your book. We are using (N=3, or triads) for each trial. However, in terms of user responses for each trial, we were slightly unsure about which among the ones below would be the correct question to ask the participant;
1. Is the first pair of stimuli more different than the second pair?
2. Which pair, 1st or 2nd is more different?

Although similar, they do have slightly different implications according to our view. For instance, if both pairs appear the same to the user, the answer to question 1 would be an unambiguous NO. However, for the second question, there is a 50% chance to say 1 or 2. Over the large number of trials that MLDS requires, we believe this would have implications on the resulting perceptual scale. So, we would love to hear your thoughts on the following;

Do you think our thinking is right in regards to (1) and (2) having different implications?
Regardless of our thinking, is there a correct question to ask while designing the experiment or does it not matter?

It's hard to say whether there would be any difference in results depending on which of the two Qs you used, but I would definitely favour Q2 as it is intuitively the more balanced of the two questions.

Part 2:
I noticed from the literature that MLDS has been applied in various vision based experiments that generally do not have to consider temporality as a factor. Eg: All four stimulus are presented in the screen at once and a participant responds based on that. However, in the context of haptic based experiments, each stimuli, even within a single trial have to be presented in a certain temporal order. One of the concerns we have is that, let's say by the time we present the second pair of stimuli and ask the question "which one is more different", participants might have somewhat forgotten about the first pair, thus introducing additional noise to the responses. With that said, I think large number of trials along with randomization helps mitigate this possible temporal bias. I would really love to hear your thoughts on this and/OR if you know of a resource that tackles this problem, it would be great if you could point me to it.

The point you raise about temporal order is a good one: as you say by the time you get to the second pair the observer might have forgotten what the first pair looked like. This is a general problem in all procedures involving a temporal order of presentation (e.g. the temporal odd-man-out task) and the best rule is always to try to minimise the cognitive load involved. Triads is possibly the best solution here, if it's possible.


best wishes

Fred

[gsharma70]

Hi Fred,

Thank you so much for your response. Really appreciate it.

For part 1, we were really not sure which way to go, but based on your recommendation, we will go for q2 and for part 2, we are planning to setup triads, which matches your recommendation. Thanks a lot.

Best,
Gyanendra