The theoretical landscape of scientific studies of consciousness has flourished. Today, even multiple versions of the same theory are sometimes available. To advance the field, these theories should be directly compared to determine which are better at predicting and explaining empirical data. Systematic inquiries of this sort are seen in many subfields in cognitive psychology and neuroscience, e.g. in working memory. Nonetheless, when we surveyed publications on consciousness research, we found that most focused on a single theory. When ‘comparisons’ happened, they were often verbal and non-systematic. This fact in itself could be a contributing reason for the lack of convergence between theories in consciousness research. In this paper, we focus on how to compare theories of consciousness to ensure that the comparisons are meaningful, e.g. whether their predictions are parallel or contrasting. We evaluate how theories are typically compared in consciousness research and related subdisciplines in cognitive psychology and neuroscience, and we provide an example of our approach. We then examine the different reasons why direct comparisons between theories are rarely seen. One possible explanation is the unique nature of the consciousness phenomenon. We conclude that the field should embrace this uniqueness, and we set out the features that a theory of consciousness should account for.

Across a century or more, ambiguous stimuli have been studied scientifically because they provide a method for studying the internal mechanisms of the brain while ensuring an unchanging external stimulus. In recent years, several studies have reported correlations between perceptual dynamics during bistable perception and particular brain characteristics such as the grey matter volume of areas in the superior parietal lobule (SPL) and the relative GABA concentration in the occipital lobe. Here, we attempt to replicate previous results using similar paradigms to those used in the studies first reporting the correlations. Using the original findings as priors for Bayesian analyses, we found strong support for the correlation between structure-from-motion percept duration and anterior SPL grey matter volume. Correlations between percept duration and other parietal areas as well as occipital GABA, however, were not directly replicated or appeared less strong than previous studies suggested. Inspection of the posterior distributions (current “best guess” based on new data given old data as prior) revealed that several original findings may reflect true relationships although no direct evidence was found in support of them in the current sample. Additionally, we found that multiple regression models based on grey matter volume at 2–3 parietal locations (but not including GABA) were the best predictors of percept duration, explaining approximately 35% of the inter-individual variance. Taken together, our results provide new estimates of correlation strengths, generally increasing confidence in the role of the aSPL while decreasing confidence in some of the other relationships.