A common neural code for similar conscious experiences in different individuals

Abstract

The interpretation of human consciousness from brain activity, without recourse to speech or action, is one of the most provoking and challenging frontiers of modern neuroscience. We asked whether there is a common neural code that underpins similar conscious experiences, which could be used to decode these experiences in the absence of behavior. To this end, we used richly evocative stimulation (an engaging movie) portraying real-world events to elicit a similar conscious experience in different people. Common neural correlates of conscious experience were quantified and related to measurable, quantitative and qualitative, executive components of the movie through two additional behavioral investigations. The movie's executive demands drove synchronized brain activity across healthy participants' frontal and parietal cortices in regions known to support executive function. Moreover, the timing of activity in these regions was predicted by participants' highly similar qualitative experience of the movie's moment-to-moment executive demands, suggesting that synchronization of activity across participants underpinned their similar experience. Thus we demonstrate, for the first time to our knowledge, that a neural index based on executive function reliably predicted every healthy individual's similar conscious experience in response to real-world events unfolding over time. This approach provided strong evidence for the conscious experience of a brain-injured patient, who had remained entirely behaviorally nonresponsive for 16 y. The patient's executive engagement and moment-to-moment perception of the movie content were highly similar to that of every healthy participant. These findings shed light on the common basis of human consciousness and enable the interpretation of conscious experience in the absence of behavior.

Keywords: brain injury; disorders of consciousness; fMRI.

Fig. 1.

Brain-wide synchronization of neural activity across subjects. (A) Movie viewing elicited significant (P < 0.05; FWE cor) cross-subject correlation across the brain. (B) No cross-subject correlation was observed in the resting state. (C) The scrambled movie elicited significant (P < 0.05; FWE cor) cross-subject correlation only within primary and association visual and auditory cortex; none was observed in higher-order, supramodal cortex. (D) The intact movie elicited significantly (P < 0.05; FWE cor) more cross-subject correlation than the scrambled movie bilaterally in parietal, temporal, motor, and dorsal/ventral frontal/prefrontal cortex. Warmer colors depict higher t values of cross-subject correlation.

Fig. 2.

Cross-subject synchronization of neural activity within different functional brain networks. (A) Time courses of group-level ICA components (ICs) clustered into five groups; they were correlated within groups and anticorrelated/less correlated across them (A, 1–5). (B–F) Individual ICs clustered into five spatially distinct brain networks (Fig. S1). (G) Individual time courses of independent components (IC) explaining the most variance within the frontoparietal network. (H) The single-subject time courses of IC explaining the most variance within each network were significantly synchronized across individuals. (I) No synchronization was observed between single-subject ICs of the resting state data, as, for example, shown by the auditory IC explaining the most variance overall in the resting state dataset. S1–12, Subject 1–12.

Fig. 3.

Decoding shared executive engagement in healthy participants. (A) Performance on the SART (30) during simultaneous movie viewing. The probability density function, to the Right, shows that SART responses followed the canonical pattern (30); i.e., responses preceding the erroneous key press (where responding was automatic; pink) were significantly faster than those preceding the correct withhold (where executive processes were engaged; blue). (B and D) Overlay of the group-level frontoparietal IC (red) and the frontoparietal activity (P < 0.05; FWE cor.; green) predicted by SART performance (B) and frame ratings (D); overlap areas are displayed in yellow. (C, Middle) Group-averaged suspense ratings of movie stills. (Top, a–f) highest-rated and (Bottom, g–l) lowest-rated frames. Higher-rated frames predicted stronger activity within the frontoparietal network (D).

Fig. 4.

Predicting sensory-driven and higher-order processes in individual participants. (A–C). Single-subject SPM analyses probe individual participants’ responses. The processing of the movie’s auditory (A), visual (B), and executive (C) information in each healthy participant was significantly (P < 0.05; FWE cor) predicted by the time course of the respective brain network in the leave-one-out group ICA. S1–S12, Subject 1–12.

Fig. 5.

Decoding executive function in one behaviorally nonresponsive patient. Healthy group: (A) Group-level auditory (purple) and visual (blue) ICs. (B and C) The healthy group’s activity predicted by the quantitative (B)/qualitative (C) executive measure (green) is overlaid on the group frontoparietal IC (red); overlap areas are displayed in yellow. Patient 1: (A) The healthy group’s auditory IC predicted significant activity in Patient 1’s auditory cortex (purple). (B and C) No evidence of visual responses or executive processing similar to the healthy participants’ was observed. Patient 2: (A) The healthy group’s auditory and visual ICs predicted significant activity in Patient 2’s auditory (purple) and visual (blue) cortex, respectively. (B and C) The quantitative (B) and qualitative (C) executive measures predicted activity (green) in the Patient’s frontal and parietal regions. Overlap with activity predicted by the healthy group’s frontoparietal IC (red) is displayed in yellow.