The limited resolution of current neuroimaging techniques poses a substantial problem for the investigation of the encoding of contents in the human brain, and thus for studies on NCCCs in the human brain. The most important format in which information is coded in the brain is the cortical column (Fujita et al. 1992). Cortical columns consist of small groupings of cells with similar tuning properties, clustered together at a scale of around half a millimetre. Even functional magnetic resonance imaging (fMRI) does not routinely have a sufficient resolution to selectively study the activation of individual cortical columns (but see e.g., Yacoub et al. 2008 for recent progress). For this reason, most research into perceptual contents has relied on experimental “tricks” that allow the tracking of contents indirectly.
In frequency tagging, a visual stimulus is tagged with a specific and unique flicker frequency. This then allows for tracing of the processing of this stimulus by searching for brain signals that exhibit the same flicker frequency. This approach has been used to study binocular rivalry, but in quite a different way to that undertaken by Leopold & Logothetis (1996). Tononi et al. (1998) tagged the inputs of the two eyes with different frequencies. They found that the currently dominant percept was accompanied by wide-spread increases in Magnetoencephalography (MEG)-signals at the tagged frequency across multiple brain regions, mostly in the early visual and temporal cortex. This is a very powerful approach and it reveals how wide-spread the effects are when a stimulus reaches visual awareness. However, it is not always clear whether these findings indicate that the corresponding perceptual features of the stimuli, in this case the orientation of line elements, really are distributed throughout the brain. The key problem is that the feature that is traced (the frequency) is not the main feature that is perceptually relevant (orientation). One could imagine, say, that activity in higher-level brain regions that exhibits the frequency of the dominant stimulus might not be involved in coding the sensory content, but instead in detecting the presence of a change in the visual image, irrespective of what the corresponding feature is. The frequency-tagging approach does not allow for distinguishing between these alternatives.
Another approach to tracking content-selective processing is to use stimuli that are known to activate specific content-selective brain regions (Tong et al. 1998; Rees et al. 2000). For example, in a study on binocular rivalry, Tong et al. (1998) used faces and houses as rivalry stimuli. These stimuli are known to activate different brain regions, the fusiform face area (FFA) and the parahippocampal place area (PPA). They found that activity in a content-selective region increased when the corresponding stimulus became perceptually dominant. This goes further than the frequency-tagging approach in that it allows for drawing the plausible conclusion that awareness leads to increased activity in content-selective regions. However, this approach again suffers from several problems. First, it only allows us to address the hypothesis related to very specific stimuli (typically faces and houses) and to very specific brain regions. Because the approach relies on the existence of macroscopic content-selective regions, it would not be possible to test whether, say, the prefrontal cortex, receives sensory information when a stimulus reaches awareness. A further problem is that the high selectivity of FFA and PPA has long been questioned (Haxby et al. 2001).