Interpreting fMRI data

[anhinga_anhinga]

Our thinking about fMRI data is too straightforward.

The abstract below is quite remarkable (my boldface).

****************************************************

(MIT) BCS COLLOQUIUM SERIES, Friday, 18 April 2008 at 4:00 PM.

Aniruddha Das, Center for Neurobiology and Behavior Columbia University, College of Physicians and Surgeons

BCS Auditorium, 46-3002

Entraining to Task Expectation in Primary Visual Cortex.

http://mit.edu/~bcs/newsevents/colloquia.shtml



Abstract:

In functional brain imaging the underlying hФmodynamic signals are assumed to reflect cortical metabolic demand driven by local neuronal responses. To test this assumption we developed a dual-wavelength optical imaging technique that simultaneously measures cerebral blood volume and blood oxygenation, continuously, in alert behaving monkeys. Using this technique, we have discovered a novel cortical blood flow signal that appears to be driven by task timing, independent of local neuronal responses. We see this signal in primary visual cortex (V1) of monkeys engaged in periodic fixation tasks, even in total darkness, with little or no measurable underlying neuronal activity. The entrainment to trial period suggests anticipatory timing in the brain.
Given a predicted event, fresh arterial blood appears to be pumped in, in anticipation of the upcoming event, before any spiking driven by the event. This signal could form part of a preparatory mechanism in the brain that is engaged by any periodic or predictable task. It could be governed by a central timing mechanism and mediated through neuromodulatory control of cortical blood flow, independent of local neuronal spiking and related metabolic demands.

Comments

[misha-b.livejournal.com]

what do you make of it?

[anhinga-anhinga.livejournal.com]

this is just a reminder to be careful in data interpretation..

typically, people assume that if they see fMRI signal, this means neurons are firing there..

since so much neuro research is based on fMRI data these days, there must be more situations like the one described here..

[misha-b.livejournal.com]

I see, the blood actually comes before the neurons start firing.
I am slightly skeptical of FMRI actually (not the procedure itself, of course , but its interpretations), it seems that often people report connections but causality is remains very unclear.

[anhinga-anhinga.livejournal.com]

It is certainly related to neural activity (in this case, the system anticipates the activity and the associated metabolic demands). But it's not always one-to-one, and there can be temporal shift both ways too..

Although, there is a heretical school of thought, which thinks that blood flow in the brain is not just metabolic, but also computational:

http://en.wikipedia.org/wiki/Hemo-Neural_Hypothesis

Causality is difficult to establish in a system which has so many recurrent circuits and feedback loops.