Hi! We would like to present a series of affective pictures. What is the best time interval between two pictures so that we might analyze skin conductance responses in the right way? We would like to change from one picture to the other as quickly as possible to maintain the interest and provide constant affective stimulation. Thus, we are mostly interested in minimal acceptable intervals. What literature guidelines would you recommend? What is your personal experience? I will be grateful for any hint. Thank you!
Dear Lukasz,
I remember the question asked when we were subject to this procedure but not the exact time. It was not more than 5 min nevertheless. Check the method section out in these papers and recommendations you find in the attached links:
http://onlinelibrary.wiley.com/doi/10.1111/j.1469-8986.1981.tb03024.x/pdf
http://www.birmingham.ac.uk/Documents/college-les/psych/saal/guide-electrodermal-activity.pdf
Article Using conditioning to elicit skin conductance responses to deception
Article Increased skin conductance responses and neural activity dur...
Article Relaxation and Executive Control Processes in Listeners: An ...
Research Skin conductance responses and neuroimaging correlates of fe...
Dear Lukasz,
Since SCR is a lagged response, responses can add up or suppress following ones and make it more difficult to decompose a specific event-related response - similar to the BOLD response in fMRI. So a rather safe thing would be to have a larger interstimulus interval, which is, however often not practical.
Similar to fMRI, it is beneficial to jitter the interstimulus interval, i.e., using different durations of ITIs (e.g., poisson distributed), which facilitates decomposition/increases estimation efficiency. By using a jitter you may be able to significantly reduce the mean ISI. Moreover, a jitter also minimizes repetitiveness, expectations of onsets, etc.
Importantly, there are also model-based analysis procedures, which are superior than traditional analysis procedures in estimating specific responses, such as PsPM (http://pspm.sourceforge.net/), formerly known as SCRalyze, which models the latent sympathetic response. On the website is also a bunch of interesting references regarding the method.
Coming back to your specific question of duration: For Ledalab, another model-based approach (PsPM/SCRalyze, however, has been shown to be superior), I remember that the authors also decomposed responses for an ISI of 4 sec, but it might also be possible to use even shorter mean ISI (e.g., when jittered). Maybe there is also specific information on the ISIs in the PsPM/SCRalyze literature.
Best, Stefan
Have you considered answering the question empirically? You could start with an interval that is unquestionably long enough to disallow any influence of readings on those that follow then. Reduce the interval in steps until responses are affected by those that preceded them, and then, of course, back off until you find a clean interval. The effort would be justified because you would have calibrated the interval to your stimuli, equipment, setting, and experimenters (some of whom might lower response threshholds), using participants similar to those you'll recruit for the real study. You would also learn something about variability in the refractory period across participants, which might be helpful in the event of unexpected findings.
Some additional studies have investigated this question examining different quantification methods with some success for ISI between 2-4 seconds, although these methods are more sophisticated (see attached)
PS: The study of Benedek et al. (2010) mentioned by C. Beau Nelson was the study I mentioned/remembered in the last paragraph of my comment regarding Ledalab and its application to ISIs for 4 sec. However, as mentioned before, I would recommend PsPM as it has been proofed to be even more efficient. In general, such model-based approaches are highly recommended.
Considering that SCR is a quite slow process I would say that you often need a quite high ISI for the signal to return to a baseline level. In all our studies we have used 25-30 seconds but I have seen some studies who have done it with maybe minimum 15 seconds. Perhaps you could use less, but I would say that would be problematic because it is not certain that the SC has returned to a baseline level. However, it is of course also problematic that participants tend to lose interest in an experiment which contains that much waiting inbetween stimuli.
Although it is beneficial for estimation/decomposition, it is not necessary that the response returns to baseline. Similar situation in fMRI, which is also a lagged response, but shorter ISIs are often used. Jittered ISIs and model-based approaches (that include a response function) increase estimation efficiency so that also considerably shorter (mean) ISIs than 25-30 sec should be possible for most kinds of research questions.
Margaret Bradley and Peter Lang have done hundreds of studies with IAPS pictures and EDA. You could check their work or even write to Margaret (she generally takes time to reply!).
Good luck
Patrick
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India