Original Contributions When more means less: a paradox BOLD response in human visual cortex Valentine L. Marcar a, *, Andrea Straessle b , Franck Girard c , Thomas Loenneker b , Ernst Martin b a Department of Neuropsychology, Institute of Psychology, University of Zu ¨rich, Zu ¨rich, Switzerland b Department of Diagnostic Imaging, Neuroradiology and Magnetic Resonance, University Children’s Hospital, Zu ¨rich, Switzerland c GE Medical Systems SA, Buc 78533, France Received 16 May 2003; accepted 7 October 2003 Abstract The predictions of the ‘Linear Transfer Model’ (LTM) have been tested only by modulating the frequency of the action potentials while keeping the size of the activated neuronal population constant. The LTM states that the blood oxygenation level-dependent contrast (BOLD) signal is directly proportional to the neuronal activity averaged over milliseconds or seconds. We examined the influence on the BOLD response, of manipulating the size of the activated neuronal population while maintaining the electrical discharge activity constant. We performed functional MR measurements on 30 awake, healthy adult volunteers (15 male and 15 female) using a flashed and reversing checkerboard. These stimuli induced the same vascular response and the same increase in the electrical discharge activity but varied in the size of the neuronal population being activated. The BOLD response measured by the extent of activation and the BOLD signal amplitude, was larger for the flashed than to the reversing checkerboard. An assessment of the local deoxyhemoglobin (HbR) concentration indicated that the neuronal activity was lower during the flashed checkerboard than the reversing checkerboard. Because the checkerboard associated with the lower neuronal activity yielded the larger number of activated voxels and the larger BOLD signal, our results run contrary to the predictions of the ‘Linear Transfer Model’ and for this reason we refer to them as paradoxical. Stimuli defined by luminance contrast or a chromatic contrast yielded identical results. We conclude that the ‘LTM’ may apply to stimuli that modulate the electrical discharge activity but not to stimuli that modulate the size of the activated neuronal population. © 2004 Elsevier Inc. All rights reserved. Keywords: Checkerboard; Blood flow; Metabolism; Hemoglobin; Action potential 1. Introduction The observation that an increase in neuronal activity in a region of the brain is followed by an increase in blood flow to that region dates back more than a century [1]. This observation has come to form the basis of modern func- tional imaging methods such as SPECT, PET or functional MRI, which rely on it to visualize sensory, perceptual and cognitive processes in the human brain. Functional MRI has established itself as the tool of choice for investigating cerebral processes. The main reason for this is that it makes use an endogenous contrast agent, deoxyhemoglobin (HbR) to localize activated brain regions. The paramagnetic prop- erty of HbR increases the magnetic susceptibility in a con- centration-dependent manner. The higher the HbR concen- tration rises, the lower the MR signal becomes. The increase in blood flow that follows an increase in neuronal activity dilutes the local concentration of HbR to such an extent that an effective increase in the MR signal is observed. This chain of events forms the basis of the blood oxygenation level dependent (BOLD) signal [2– 6]. There is good agreement between the functional imaging methods in their ability to localize activated brain regions. A difference between them emerges when it comes to assess- ing the level of the neuronal activation. In functional MRI, a great deal of effort has been directed towards establishing the relationship between the BOLD response and the underlying neuronal activity. Such a relationship would provide a quanti- tative measure of the ongoing neuronal activity. Following the traditions of single-cell electrophysiology, studies examining the relationship between the BOLD re- sponse and the neuronal activity have modulated the action potential frequency of neurons. In striate cortex this was done by varying either the luminance intensity of the stimulus [7–13] or the temporal frequency of the stimulus [5,14]. * Corresponding author. Tel.: +41-1-634-1572; fax: +41-1-634-4342. E-mail address: v.marcar@psychologie.unizh.ch (V. L. Marcar). Magnetic Resonance Imaging 22 (2004) 441– 450 0730-725X/04/$ – see front matter © 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.mri.2004.01.019