Thank you for your time and effort.
In an operant conditioning experiment where a animal is presented with stimuli on either the left or the right side for around 150 trials a day, is it OK to use a truly random series or should one control for the number of consecutive left or right presentations? I know that in the long run they will be fairly equally distributed, but I am worried that long consecutive runs of the same side will disturb my experiment. Especially when these occur early during training. Should I use a Gellerman series like approach?
best
Nils
Hi, Nils. I have the same intuition as you that long consecutive runs of the same type of trial could disturb the outcome of a training treatment. I'm used to divide my training sessions in 8 blocks of four trials each, each block consisting in a random selection of two "A" trials and two "B" trials without replacement. This warrants to have exactly 50% of each type of trial within a session and a maximum consecutive number of 4 trials of the same type that will never occur neither at the start nor at the end of the session. Now, you have 150 trials, which is not divisible by 4, so you might use 25 blocks of 6 trials (three "A" trials and three "B" trials randomly selected, without replacement); the maximum consecutive number of trials of the same type would be 6 in the middle of the session and 3 in the first or in the last trials. Another alternative is to reduce your amount of trials per session from 150 to 148, which would allow you to have 37 blocks of 4 trials (two "A" trials and two "B" trials). I hope this would help.
In our lab, we do just what Rodrigo describes. If you use MED Notation to program the boxes, use the RANDD command (randomly select without replacement) on an array that contains the alternatives.
Federico's suggestion is good, but leaves open how to populate the array. The kinds of restriction that Rodrigo describes would probably work well. In the past, I have usually generated a random series and adjusted it afterwards. Another possibility, if you want to do something new, would be to explore "red noise" (as opposed to white noise).
- Badges
- Science topic
- Similar topics
- Methodology
- Research Methodology
- Research Papers