The problem: We are interested in how ketamine modulates hedonic experiences such as chills. Participants listen to music while they are in the scanner both on ketamine and placebo. Participants also pre-rate all the songs outside the scanner a week before. For our analysis we need to know when participants experience their peak emotional moment during the scan session.
Our ‘solutions’ that did not work: 1. Live rating – people press a button while they experience their peak moment. Problem: we confound our neuronal signal of interest (pleasure) with motor activity. 2. Additionally rating the music after the scan session. Problem: even though, we think that the peak for most songs won’t shift between the baseline rating and the in-scanner rating sessions, it might do so during the ketamine condition (makes everything more pleasurable and maybe even earlier). So, if they rate the music again afterwards when most of ketamine's effects have already subsided, we might not find the same peak moments as during the scan session. (plus, we also doubt that participants could reliably remember when those peaks occurred during the scan session…)
3. measure physiological responses. problem: yes, skin conductance does correlate with chills but we do not have the equipment for that...
Does anyone of you have an idea how we could measure the peak moment during the scan session without majorly confounding our measurements? I would really appreciate your help!
Sounds like you've got a cool study planned! I have a few ideas that may be helpful.
1. Create a deconvolution filter for the motor response or use independent components analysis to pull it out of the signal. This paper might be useful for that: Article Deconvolution of impulse response in event-related BOLD fMRI
2. If you have an MR-compatible eye-tracker, you could try measuring pupil dilation like in this study: Article Music chills: The eye pupil as a mirror to music’s soul
. The problem here though is that there didn't appear to be a difference between chills-inducing and control songs when subjects weren't responding and then there's the issue of ketamine potentially interfering with the pupil dilation response itself which is a possible confound.3. Contrast the peak emotional live rating with some other type of live rating unrelated to pleasure. This way, the cognitive demands related to motor planning and execution are held constant for both conditions so any difference will likely be related to pleasure. You could probably pick anything related to a subjective feeling or even something related to the music.
4. I don't know if you're only using chills-inducing songs or if you have control songs that won't elicit chills, but I believe that Robert Zatorre and Scott Bannister have found that chills (and the moments where they peak) are pretty reliable and consistent. Unless you really care about the temporal resolution of this peak latency, then you could probably get away with just analyzing the window a few seconds long surrounding the emotional peaks they provided you with outside the scanner beforehand. I'm not sure how much earlier you expect ketamine to shift these peaks though.
What sorts of analyses/statistics are you planning to run on this data? Are you comparing the averages of entire trials with placebo and ketamine or are you planning on looking at the time series?
Thank you for your detailed response, Connor!
Yes, we also have a neutral music condition. Similarly to the pleasure music condition, the music is also self-selected but should neither elicit negative nor positive emotions.
What we want to do in a nutshell:
easy stuff
1. simple contrast between songs that elicit & don't elicit chills
2. simple contrast between aesthetically moving and not moving songs
more complex stuff
3. look at complexity measures (e.g. LZC) and check if peak hedonic experiences correlate with peak moments of complexity
4. ketamine's effect on network modularity
ad 1.
Honestly, I am quite new to task-based fMRI research, therefore I am not completely sure if I understand this correctly. Do you suggest to model the button press (e.g. let each participant press the button a few times) and include the modeled motor response as a regressor in the model? Would you mind to elaborate on this a bit?
ad2.
Unfortunately, we do not have an eye tracker available...
ad 3.
The task is already at its max length, therefore we cannot include anything more. (Otherwise we do not have enough time points for the complexity analysis)
ad 4.
Thank you for pointing that out. Yes, I know that the peak moment stays quite stable over time, yet the problem is, we also have a neutral condition. Since we hypothesize that ketamine increases the hedonic experience, we also suspect that some previously 'neutral' rated songs will become aesthetically moving (or even might elicit chills). If this is the case, we do not have any previous accounts of where the peak moments could be...
Well the easy stuff should already be quite attainable with your current design without a live rating so if all else fails, you can probably publish based on this alone.
As for Goal 3, I think that's super interesting although I would hypothesize that peak hedonic experiences actually arise at moderate complexity given the groove literature that I work with. However, that work isn't usually calculated moment-to-moment with the exception of Marcus Pearce's stuff done with IDyOM. Related to this and Goal 4, how long are your trials?
ad 1. I really don't have a whole lot of experience working with fMRI or deconvolution, but yeah what I had in mind is some sort of functional localizer where subjects spent a handful of trials just pressing the button that you could "subtract out" of the signal in the music trials later. In EEG, it's common practice to run an independent components analysis (ICA) as part of data cleaning to find blinks because they're huge signals that stick out in a very stereotyped manner. You then remove the blink components from the signal and run an inverse ICA to get the clean signal without the blinks. Again, I don't do fMRI so I'm not sure how feasible it is (except that ICA is used for cleaning head movements which show up as "ring" artifacts), but if you ran a functional localizer you would probably have a good idea of what the movement component would look like in the ICA if the literature doesn't already have an answer.
ad 4. While you may not know when the hedonic experience/chills will occur in the context of ketamine, the literature definitely has answers on where these things occur. For example, you could potentially just take parts of the basal ganglia like the nucleus accumbens as a region of interest and just compare peak activity there as a proxy for what you're after. A post-trial rating could just be used to validate this and you can assume the peak latency in your ROI corresponds to the (time-lagged) moments of maximal hedonic experience.
Sorry, for the long delay!
Our music excerpts are 70 seconds long and we have 20 songs altogether (10 neutral/10 positive).
Would you mind sharing this groove literature with me? I would be really interested in reading papers showing this connection between moderate music complexity and peak moments!
ad 1. Yes, we thought about that but many scientist I talked to argued, that the brain's network activity is so complex and sophisticated that one cannot (should not) simply subtract this confound from the signal... But I will look into it again! Thank you!
ad 4. Mhm this is a very good idea! I will look into that aswell.
Thank you for your responses Connor!