Serendip is an independent site partnering with faculty at multiple colleges and universities around the world. Happy exploring!

Imaging And The Question of Consciousness - Paper

rdanfort's picture


Discussion Paper – Imaging and Consciousness

            Pick up a newspaper, and you will discover an amazing thing: we are living in the future. The for-real future of jet-packs, ray-guns, and mind reading. Our revolutionary imaging technology can determine what memories you are accessing, what shapes you are seeing, the degree to which a decade of happy hour has smoothed your prefrontal cortex, and whether or not you associate the face of John Edwards with a particularly debauched collection of short stories. So it would seem, anyway.

            To talk about what makes imaging interesting for me is to talk about what makes psychology interesting to me. My familiarity with the subject grew throughout high school, as I read and followed generations of scientists asking what’s going on in the brain. I was entranced first by accounts from Shulgin and Hofmann, among other early researchers, of radical drug-induced changes in perception and consciousness, these stories of cavalier experimentation evoking a “Wild West” of scientific inquiry. I took AP Psychology some few years after fMRI had really taken off, and was taken in by the crisp, brilliant patterns and promised connections to the nature of human behavior and human experience. Never before had we had this kind of access to the brain in real time, and the answers to our most profound questions might arrive shortly.

            However, just as love for the wild research of the 50s and 60s might be tempered by a better understanding of the scientific process - never mind Ewen Cameron and the torture cells of South America - the promise of fMRI does not quite meet critical examination. This was first made clear to me by Wendy Sternberg in a Sophomore-year Psychology class, but has been a growing understanding in the field for some time. It is with this in mind that Jeremy and I hoped to explore the ways in which fMRI is used and its real implications regarding brain study.

            The basic features of the gulf dividing fMRI promise from what fMRI delivers were outlined in a 2005 Scientific American article, “Fact or Phrenology?” (Dobbs, 2005.) Our compulsion to map various activities to discrete brain regions based on fMRI data is eerily reminiscent of the now-discredited mapping of skull bumps to personality traits. While the interpretation and collection of imaging data is rarely so cretinous as that of phrenologists, the article describes specific ways in which fMRI is limited. First, the time of data collection – around 60 seconds – is insufficient to capture individual neural firings. The entire measurement presumes a direct correlation between neural firing and blood oxygen levels; while such a relation exists, it is mediated by glial cells and is an order of magnitude slower than neural firing. The raw data is transformed into voxels that can contain millions of individual neurons via a questionable set of algorithms, neglecting potentially crucial activity by smaller numbers of cells. The fMRI process does not tell us about the path of a signal. It does not discriminate between excitatory and inhibitory firing. And so on.

            These limitations seriously affect the veracity of claims made using fMRI data. Without knowing where a signal originates, BOLD topography only tells us that arrival of a massive number of signals at a region correlate to a behavior. This failing of BOLD might be ameliorated by increasingly well-resolved collection that captures junctions of fewer signals, but one could imagine many configurations that would not be detectible and the prospect of isolating small changes with BOLD is technically daunting. It may be that most synapses do not initiate enough glial prostaglandin release to dilate blood vessels in the slightest. The lack of excitation/inhibition discrimination might be less of a problem if downstream events could be analyzed, but that brings us to the same question of resolution. For these and other reasons, most readers would agree that use of fMRI for certain purposes is analogous to phrenology (Grobstein, 2010.) It is agreed that well-defined studies make good use of fMRI, but the investigator and the interpreter must be clear as to the limitations of the technology when making any causative claims. 

            Among the many concerns raised about BOLD technology, the potential of current algorithmic processing for error was concisely illustrated in 2009 poster by researchers at Dartmouth, “Neural correlates of interspecies perspective taking in the post-mortem Atlantic Salmon: An argument for multiple comparisons correction” (Bennett et al., 2009.) Researchers were able to obtain fMRI readings in a dead fish presented with images and asked to describe its emotional state. The data do not suggest that the dead fish has opinions, but rather that the fMRI technique is somewhat facile. The paper’s authors have argued that fluctuations in the data are such that a stricter probabilistic threshold – using “multiple comparisons correction” was necessary to prevent recording of false positives.  At this point, some might pine for the clarity of lesion studies, which provided us with seemingly unimpeachable causation: an area is injured, and some function is lost. It bears mentioning, however, that these studies are often “grandfathered” into credibility, they suffer from the same problems that plague fMRI and any attempt to constrain complex functions to discrete regions, and they have comically small sample sizes. Blindsight, for example, did some violence to the localization of sight processing to the visual cortex (Grobstein, 2010.) This publication is especially useful because it informs both responsible fMRI users of additional ways to improve their carefully-delineated study, and may sufficiently humiliate journalists and hacks who are overly eager to explain complicated variations in behavior and thought with data that are, at best, a rough correlation. Some would go as far as say that the junction between the imaging and mechanism of mind is altogether orthogonal.

            A more fundamental problem presents itself in weighing our expectations of brain imaging, that of determining whether the basic activity of neurons correlates in a meaningful way to our personal experience. Nicholas Humphrey comes down strongly on the negative side of this argument, in his essay, “How to solve the mind-body problem” (Humphrey, 2000.) He argues that imaging is unlikely to tell us the function of these neurons or regions, just as a SEM image of a microprocessor will not tell you what programs or assembly operations a computer is running. This is important, because when we explore consciousness, we are not only asking what physical mechanisms produce it, but how they produce it. 

            Neuroscience has attempted to do some of this in dividing the brain into functional regions, analyzing the apparent behavioral roles of certain neurotransmitters, etc., but we have yet to enunciate a theory of how these pieces behave in such a way that is equivalent to the qualia, self-reflection, and other features the system is known to be capable of. Humphrey’s attempt to address the particular question of how qualia might function is largely philosophical, although informed by a “story” of natural selection. He imagines sensation as a sort of “facial expression”, one made by the brain (rather than sent to the brain) in response to stimulus. This mechanism might come about as single-celled organisms, which have analogous direct and local responses to stimuli (i.e. chemotaxis,) evolve to form differentiated multi-cellular organisms. As such a process completes, it is less adaptive for every cell to process the entire organismal response to a stimulus – it becomes advantageous for systems to form which centralize and privatize cell stimulus. 

            This is the primordial brain – a central structure that re-produces distal input and directs response based on that re-production. Increased competition would drive further differentiation and require that reproductions – sensations – become more complex, as the response – perception, thought – increases similarly in nuance and robustness. If we agree with Humprhey’s story, sensation is not an input of consciousness, but an integral function of consciousness. This is easily misunderstood as drawing equivalence between sensation and self-awareness, but can be clarified in terms of separating out the chronological phases of consciousness (Grobstein, 2010.) In analyzing sensation, Humphrey speaks to “immediacy,” the present, phenomenal consciousness; further theorizing may integrate this new idea into a broader framework for consciousness that can be tested and refined, giving us a better idea of what the neurons and synapses are doing.

            This information is likely to sober a discussion of applied neuroscience, particularly in law, security, interrogation, and advertising. Few who read these selections would agree that “detecting lies” and “reading minds” in a classical sense is feasible, and recent talk of allowing fMRI data into court to those ends is not supportable to the so enlightened (Grobstein, 2010.) It is likely that a technology capable of deciphering low-level neuronal processes would need to be calibrated to the individual for an extremely long period of time, or otherwise constrained by the wild variation in individual construction of meaning and logic. Despite that, it is distantly conceivable that “pre-verbal” thought content might be both accessible and intelligible through imaging, as it is part of a process we ourselves use to convey meaning between differently-structured individuals. However, it is unlikely that this would really tell you more than an individual would willingly divulge in conversation. Even accessing this content would require such an advance in our understanding of brain function that lie detection would be the least of the emergent controversies and problems.

Works Cited

Dobbs, D. (2005, April/May). Fact or phrenology?. Scientific American, 16, Retrieved from doi: 10.1038/scientificamericanmind0405-24

Bennett, CM. (2009). Neural correlates of interspecies perspective taking in the post-mortem atlantic salmon: an argument for multiple comparisns correction. Proceedings of the Organization for Human Brain Mapping Abstracts,

Grobstein, P. (2010, February, 27). Neural and Behavioral Sciences Senior Seminar: Imaging and the Question of Consciousness [Discussion and Online Response]. Retrieved from /exchange/courses/nbs/s10/home

Humphrey, N. (2000). How to Solve the mind-body problem. Journal of Consciousness Studies, 7. Retrieved from



Paul Grobstein's picture

imaging and applied neuroscience

" This information is likely to sober a discussion of applied neuroscience, particularly in law, security, interrogation, and advertising." 

I would like it to be so, but am perhaps less certain that information is actually by itself an effective sobering agent.