tag:blogger.com,1999:blog-5427587583898590293.post7927558377781894617..comments2023-10-31T03:41:34.562-07:00Comments on Brad Buchsbaum's Blog: fMRI is not an inherently correlational methodBrad Buchsbaumhttp://www.blogger.com/profile/10757537675625801119noreply@blogger.comBlogger10125tag:blogger.com,1999:blog-5427587583898590293.post-87851905812273704792021-05-19T02:59:28.748-07:002021-05-19T02:59:28.748-07:00Excellent post. I want to thank you for this infor...Excellent post. I want to thank you for this informative read, I really appreciate sharing this great post. Keep up your work…<br /><a href="https://ralphitness.com/programs/self-discovery-session/" rel="nofollow">transcranial magnetic stimulation therapy</a><br />Joe Roothttps://www.blogger.com/profile/10935524442959906818noreply@blogger.comtag:blogger.com,1999:blog-5427587583898590293.post-7358119573221844102012-01-08T13:25:28.115-08:002012-01-08T13:25:28.115-08:00I was responding more to the comment than to your ...I was responding more to the comment than to your article itself...I agree with you and like your post.MadScientisthttps://www.blogger.com/profile/17026578185908907713noreply@blogger.comtag:blogger.com,1999:blog-5427587583898590293.post-19624997197241708132012-01-08T13:19:24.595-08:002012-01-08T13:19:24.595-08:00@MadScientist,
It's true that the machine its...@MadScientist,<br /><br />It's true that the machine itself is not the physical cause of the intervention, but it's a distinction without a difference as it relates to the ability to make a causal inferences.Brad Buchsbaumhttps://www.blogger.com/profile/10757537675625801119noreply@blogger.comtag:blogger.com,1999:blog-5427587583898590293.post-86825162031354212252012-01-08T11:54:23.820-08:002012-01-08T11:54:23.820-08:00fMRI, like EEG, is not the means of the interventi...fMRI, like EEG, is not the means of the intervention, it is used to measure the effects of interventions (task instructions, drugs, what have you). TMS, on the other hand, is not used to make measurements, but to cause an intervention (by itself, or along with others) whose effects are then measured separately (reaction time, or perceptual accuracy, or even fMRI/EEG).MadScientisthttps://www.blogger.com/profile/17026578185908907713noreply@blogger.comtag:blogger.com,1999:blog-5427587583898590293.post-78513615370407925772012-01-08T11:46:48.735-08:002012-01-08T11:46:48.735-08:00@MadScientist
you'll have to define "int...@MadScientist<br /><br />you'll have to define "interventional technique" for me. FMRI frequently involves an experimental intervention.Brad Buchsbaumhttps://www.blogger.com/profile/10757537675625801119noreply@blogger.comtag:blogger.com,1999:blog-5427587583898590293.post-90298494092779476962012-01-07T18:41:11.101-08:002012-01-07T18:41:11.101-08:00Isn't it simply the case that TMS is an interv...Isn't it simply the case that TMS is an interventional technique, while fMRI is a measurement technique ? What am I missing ?MadScientisthttps://www.blogger.com/profile/17026578185908907713noreply@blogger.comtag:blogger.com,1999:blog-5427587583898590293.post-21801806813819074422012-01-07T18:39:58.572-08:002012-01-07T18:39:58.572-08:00This comment has been removed by the author.MadScientisthttps://www.blogger.com/profile/17026578185908907713noreply@blogger.comtag:blogger.com,1999:blog-5427587583898590293.post-57067439035130230342011-03-22T15:57:40.906-07:002011-03-22T15:57:40.906-07:00I agree that strictly speaking fMRI is not correla...I agree that strictly speaking fMRI is not correlational. If we take a behavioral point of view however I do think that fMRI can not tell us anything but correlations. A behavior scientists is not interested in which brains areas are activated by a cognitive task, but in which brain ares produce certain behavior. In other words, from the point of view of a behavioral science, a research method should use a dependent variable which measures behavior manifestations (RT, accuracy rate, eye fixations, test scores, etc) in order to reveal meaningful causal relationships. To put it simply, behavioral sciences are interested only in the brain->behavior direction of the possible causal relations between these two entities. Perhaps brain scientists are interested in the other direction as well. I personally cannot understand why it would be interesting to know that a cognitive process activates certain brain areas if we assume that the reverse is not true. It is only interesting if we assume that the relation is bidirectional and we cannot make such conclusions using fMRI.Ivan Vankovhttps://www.blogger.com/profile/06460091384573130861noreply@blogger.comtag:blogger.com,1999:blog-5427587583898590293.post-83609549793794008432010-01-05T19:45:30.582-08:002010-01-05T19:45:30.582-08:00Hi Jasper,
thanks for the comment.
You say: &quo...Hi Jasper,<br /><br />thanks for the comment.<br /><br />You say: "fMRI is correlational. Task modifications are causal". If we view fMRI as a way of measuring the brain over time -- a dependent variable -- it doesn't really make sense to state categorically that it is "correlational" -- correlational with what? It takes two variables to tango.<br /><br />Moreover, if task modifications (as you say) are causal, then task modifications with fMRI are also causal.<br /><br />It's true that changes in activation might be results of changed task, changed strategy,subject fatigue, etc. -- but this has nothing to do with fMRI per se, this is true of all psychological experiments, including reaction time, eye-tracking, EEG, TMS, and so on. Nobody, however, says that reaction time experiments are inherently correlational.<br /><br />The factors you refer to are mitigated by randomization of conditions and other methods of experimental control. True causal inference requires experimental control, of course, but this is a concern for all types of experiments, not just fMRI.<br /><br />so, while it was a nice effort, you have not succeeded in persuading me that fMRI is a fundamentally correlational method!<br /><br />BradBrad Buchsbaumhttps://www.blogger.com/profile/10757537675625801119noreply@blogger.comtag:blogger.com,1999:blog-5427587583898590293.post-81966106473552501612009-12-22T04:37:26.491-08:002009-12-22T04:37:26.491-08:00fMRI is correlational. Task modifications are caus...fMRI is correlational. Task modifications are causal. Brain changes that correlate with causal task modifications are interpreted as the result of said changes. Or the result of the result of said changes.<br /><br />However, you cannot tell whether the changes in activation you find are the result of your changed task, changed strategy, changed response, or changes in non-task activation (dozing off for half the task instead of one third, because it is more boring). <br /><br />With TMS, if you apply task manipulation A, then observe activation B, disrupt area Y, and no longer find behavior Z (but still B), it will inform you that Y is not required for B, but is required for Z. If you found neither activation B nor behavior Z, this would mean that Y -> BZ.<br /><br />Of course you can make causal inferences in fMRI using clever designs, and with analyses such as Dynamic Causal Modelling, but that does not render the method itself any less correlational. Your task manipulation plays a causal role, and perhaps the neural activity that you also observe does, but the fMRI does not.<br /><br />This is where TMS differs. Your task does not have to actively cause changes (though it can), the method itself does.<br /><br />Still, you are correct in saying that you can infer causal relationships using carefully designed fMRI experiments and analyses, and you can do TMS experiment that will not shed any light on causation.<br /><br />Basically, TMS allows you to tell that the region you stimulate is ‘somewhere upstream’ from the behavioral difference you find, whereas fMRI allows you to tell that the region you measure is ‘somewhere downstream’ of the difference you cause.Jasper Winkelhttps://www.blogger.com/profile/14846935812815591130noreply@blogger.com