Acute or transient pain activates a large set of brain regions, including thalamus, primary and secondary somatosensory areas, insula, anterior cingulate cortex (ACC), and periaqueductal gray matter, areas collectively referred to as the pain matrix, with variable activation of striatum (dorsal and ventral), amygdala, and medial and dorsolateral prefrontal cortex (Figure).1 Activity in areas of the pain matrix has been consistently observed in hundreds of studies irrespective of the modality used to elicit nociceptive input (eg, thermal heat, painful pressure, intramuscular injections, visceral balloon inflation), leading some investigators to propose that this matrix of regions mediates the conscious perception of pain.2 In addition to the consistency of activation across studies, proponents of the pain matrix advance the argument that subjective reports of pain intensity correlate with neural response magnitude within some areas of the pain matrix, mainly insula and ACC, and suggest as a corollary that modulation of pain experience leads to a corresponding modulation of the neural response within the pain matrix. More recently, Wager et al3 added strength to these arguments by showing, using a machine-learning approach, that a weighted pattern of brain activity within the areas presented in the Figure in red can be derived from one group of individuals and used to predict pain intensity in a new group or can differentiate somatosensory pain from nonnoxious thermal heat or the pain of social rejection. The pain matrix has therefore been considered sufficient to generate the conscious perception of pain elicited by peripheral nociceptive input via A delta and C fibers. However, serious challenges have been raised against this view. In a series of experiments using nociceptive, somatosensory, auditory, and visual stimuli, Mouraux et al4 demonstrated that activation within the pain matrix is multimodal rather than pain specific, and they suggested that the magnitude of neural activation can be explained by the saliency of the stimulus independent of modality. Using evoked potentials, the same authors had previously demonstrated how the context of nociceptive stimulus presentation, eg, comparing repetitive monotonous stimuli vs novel stimuli, can decorrelate the brain response in ACC and insula from the perceived pain intensity, providing counterexamples to the concept of the pain matrix.5 More recently, they showed that patterns of brain activity elicited in response to nociceptive, visual, tactile, or auditory stimuli derived from the noncorresponding primary sensory cortex, eg, response to pain derived from the auditory cortex or the visual cortex, is sufficient to differentiate between different types of peripheral input, including pain.6 These results suggest that even the traditional view of primary sensory cortices, let alone large-scale matrices like the pain matrix, being specialized in processing information exclusively from 1 sensory modality has to be abandoned for a more multisensory or multimodal view.