… the single most critical piece of equipment is still the researcher's own brain. All the equipment in the world will not help us if we do not know how to use it properly, which requires more than just knowing how to operate it. Aristotle would not necessarily have been more profound had he owned a laptop and known how to program. What is badly needed now, with all these scanners whirring away, is an understanding of exactly what we are observing, and seeing, and measuring, and wondering about.

—Endel Tulving, interview in Cognitive Neuroscience (Gazzaniga, Ivry, & Mangun eds. [New York: Norton, 2002], 323)

Introduction

With the advent of positron emission tomography (PET) in the 1980s (Fox et al., 1986) and functional magnetic resonance imaging (fMRI) in the 1990s (Kwong et al., 1992; Ogawa et al., 1992), neuroimaging became a keystone for the growing field of cognitive neuroscience. Historically, our knowledge of the human brain has been far more limited than for brains of other species (Crick & Jones, 1993), primarily due to the restriction against using invasive techniques, such as single neuron recording, in humans. Although human neuropsychological studies have been very enlightening, newer noninvasive neuroimaging methods, particularly fMRI, enable exploration of the normal rather than the disordered human brain and allow resolution at a fine spatial scale that lesions rarely provide (Savoy, 2001). These new imaging methods have identified dozens of functionally specific areas in the human brain, some of which seem comparable with areas in other species, particularly the macaque monkey, and some of which may be uniquely human (Culham & Kanwisher, 2001; Duncan & Owen, 2000; Grill-Spector & Malach, 2004; Tootell et al., 1998). Within numerous human regions that have been identified, detailed explorations have revealed the underlying computational processes (e.g., Wandell, 1999).

The growth of neuroimaging has been phenomenal. Neuroimaging publications continue to increase exponentially (Fox, 1997), with recent estimates of four published papers each day (Tootell et al., 2003). The caliber of imaging papers has improved considerably with the growth of the field, primarily due to the theory-driven approaches and high standards of experimental design expected in other disciplines. However, for many newcomers, fMRI methodology can seem overwhelmingly complex, leading to an increasing demand for resources to learn neuroimaging techniques. Numerous resources have suggested how neuroimaging experiments should be designed, but show limited consideration of the philosophy behind experimental design. One exception is an article by Steve Kosslyn, whose clever title posed the question “If neuroimaging is the answer, what is the question?” (Kosslyn, 1999). Another exception is William Uttal's book The New Phrenology (Uttal, 2001), which poses numerous pessimistic criticisms of the entire brain imaging enterprise (for a rebuttal, see Donaldson, 2004).

Here I intend to present a brief overview of neuroimaging design principles, followed by a somewhat opinionated review of my thoughts regarding the types of questions for which neuroimaging is (and is not) suited. As well, I will outline some of the principles that can lead to better questions and better experimental designs. The emphasis will be on fMRI, but many of the same principles apply to studies performed with PET. Space does not permit detailed recommendations for design issues, but the reader will find many useful resources in print (Aguirre & D’Esposito, 1999; Buckner & Logan, 2001; Chein & Schneider, 2003; Huettel et al., 2004; Jezzard et al., 2001) and online (such as my Web site, fMRI for Dummies, http://defiant.ssc.uwo.ca/Jody_web/fmri4dummies.htm). By no means do I intend to present a cynical view of the brain imaging enterprise. Rather, I hope to prompt newcomers to the field to think carefully about their approaches and the caveats, so that their contribution to the enterprise will become more fruitful.