26 September 2008Introduction to CognitionImaging of Attention and Perception in Human Early Visual Cortex – RessUsing fMRI in conjunction with a contrast detection experiment, Ress and colleagues demonstrate similarities between the retinotopy of attention and vision in the brain. By comparing the neural response to attentional stimuli and veridical stimuli, Ress demonstrated “stimulus-independent” activity in the visual cortex, finding a large amplitude fMRI response independent of the presentation of a stimulus pattern, indicating that attention alone was sufficient to trigger cortical neural activity. Further, this activity was highly related to subsequent performance measures, corroborating the hypothesis that such activation was indeed indicative of attention. Ress’ images were compelling and his use of flattened retinotopic maps of the visual cortex were exceedingly illuminating and certainly the notion of the link between attention and perception fascinating. However, because the use of fMRI techniques is so central to his research, I’d like to focus on discussing some of the questions and issues that seem to be raised by such methods and conclude with some thoughts regardingimaging of attention in the visual cortex. Ress provides an overview of the fMRI techniques used in his studies of visual attention. BOLD (Blood Oxygen Level Dependent) fMRI is an indirect method of measuring cortical neural activity via the hemodynamics of blood flow and oxygenation in the brain. Such a technique, which detects the magnetic signal differential between oxygenated (non-magnetic) and deoxygenated (paramagnetic) blood, is predicated on the assumption that blood flow is a reliable, linear proxy for brain activity because active neurons consume the oxygen component of hemoglobin, causing higher rates of oxygen release to active areas. Although I am unfamiliar with the research myself, Ress argues that a burgeoning corpus of evidence has corroborated the assumption correlating blood flow with neural activity, although the exact mechanisms of their relationship seems poorly understood. This critical issue aside, BOLD fMRI techniques seem to have a number of other limitations and assumptions which must be considered in interpreting such research, and which are especially relevant in light of the profusion of recent fMRI studies. First, as Ress notes, hemodynamic studies have revealed that increased blood flow to active regions occurs after a hemodynamic lag of four to six seconds. Because of this lag, BOLD fMRI affords researchers relatively poor temporal resolution in the context of high spatial resolution. This may be contrasted with a technique such as electroencephalography (EEG), which has high temporal resolution, but relatively poor spatial resolution owing to the placement of electrodes on the scalp (discussed in greater detail below). Interpretation of fMRI results thus seems to assume a linear and proportional, albeit temporally shifted, relationship between hemodynamic response and neural activity. It seems possible, however, that these responses are not, in fact, proportional to actual neural activity. Further, while such techniques assume that blood flow is a response to neural activity and oxygen consumption, it seems feasible that blood flow may in fact precede neural activity insome cases, reflecting anticipation of or preparation for future neural activity. Further, while fMRI may be a good correlate of “neural activity”, it is unclear exactly what “neural activity” is. Does neural activity refer only to excitatory activity, or also inhibitory? Does it refer only to cell action potentials, or to increased subthreshold activity? Does it refer to functional activity (i.e., areas that are processing the stimulus), or to more peripheral or second-order activities in ancillary cortical areas not directly related to the processing of the stimulus? In addition, the primary advantage of fMRI research seems to be localization of brain activity,which seems somewhat (although certainly not entirely) inconsistent with the emerging connectionist emphasis on parallel distributed networks of cortical activity, which emphasizesthe parallel and distributive nature of neural processing and representations. In addition, whileBOLD fMRI seems especially useful for localizing neural activity in response to certain stimuli and visualizing the waves of neural activity, it provides little information on how thoseareas are responding. As discussed in Trujillo’s presentation, alternative techniques such as electroencephalography (EEG), which measures neural activity directly via the measurement of electrical activity via electrodes on the scalp, may provide complementary information with better temporal resolution than fMRI (albeit, in the case of EEG, with lower spatial resolution). Such imaging techniques, therefore, provide complementary information regarding neural activity, inasmuchas they provide high-resolution spatial and temporal information regarding neural activity via indirect hemodynamic and direct electrical imaging, respectively. Certainly, although I am sure such research has been conducted to the extent permitted by such technologies, it would be interesting to see research similar to Ress’ conducted via the concurrent use of fMRI and EEG technique. Although neither Ress not Trujillo mentioned it, I am under the impression that magnetoencephalography (MEG) imaging, which detect the magnetic properties of electrical brain activity directly, also seems like a promising route minimizing some of the limitations of traditional fMRI techniques, providing relatively high spatial and temporal resolution. Regarding the issue of detection of attention via fMRI, given the critique above, that it is difficult to interpret what the neural activity detected by the fMRI actually means because the fMRI can only identify patterns of activation, and not reveal what the underlying cells are doing, it is unclear how attention differs from perception at the neural level. How does attention to a stimulus differ from actually perceiving a stimulus? How does attention differ from mental imaging (in the absence of a veridical stimulus) of an object? How could fMRI or other imaging techniques be designed to detect such a difference? Certainly, central to suchquestions is the very concept of attention
View Full Document
Unlocking...