The neuroscience of advanced scientific concepts

Abstract

Cognitive neuroscience methods can identify the fMRI-measured neural representation of familiar individual concepts, such as apple, and decompose them into meaningful neural and semantic components. This approach was applied here to determine the neural representations and underlying dimensions of representation of far more abstract physics concepts related to matter and energy, such as fermion and dark matter, in the brains of 10 Carnegie Mellon physics faculty members who thought about the main properties of each of the concepts. One novel dimension coded the measurability vs. immeasurability of a concept. Another novel dimension of representation evoked particularly by post-classical concepts was associated with four types of cognitive processes, each linked to particular brain regions: (1) Reasoning about intangibles, taking into account their separation from direct experience and observability; (2) Assessing consilience with other, firmer knowledge; (3) Causal reasoning about relations that are not apparent or observable; and (4) Knowledge management of a large knowledge organization consisting of a multi-level structure of other concepts. Two other underlying dimensions, previously found in physics students, periodicity, and mathematical formulation, were also present in this faculty sample. The data were analyzed using factor analysis of stably responding voxels, a Gaussian-naïve Bayes machine-learning classification of the activation patterns associated with each concept, and a regression model that predicted activation patterns associated with each concept based on independent ratings of the dimensions of the concepts. The findings indicate that the human brain systematically organizes novel scientific concepts in terms of new dimensions of neural representation.

Fig. 1. The fMRI-based factor scores for the 45 concepts on the measureable magnitude dimension (vertical axis).

The contrast between the concepts at the two ends of the dimension (highlighted in blue text) indicates the nature of the dimension.

Fig. 2. Factor scores for the measurable magnitude factor from the fMRI brain data (x-axis) were well correlated (r = 0.74) with the mean expert ratings of the concepts with respect to that factor (y-axis).

The concepts at the extreme ends of the factor score distribution that are influential for factor interpretation are highlighted in blue in a larger font.

Fig. 3. The predictive model presented graphically and as data.

The left panel is a schematic representation of the predictive model. The right panel shows a scatterplot of observed and predicted activation values in the 30 factor-related clusters for a sample concept, dark matter, where R² = 0.85. For this illustration, the predictive model was applied to a mean dataset obtained by averaging the activation data of all participants, and developing the mapping from the ratings of the other 44 concepts along the four main factors to the mean activation level of the 30 cluster locations associated with the factors. The resulting regression weights were then applied to the ratings for the left-out (45th) concept (dark matter) to predict its activation values in those 30 locations.

Fig. 4. Factors locations associated with the post-classical end of the classical vs. post-classical dimension (i.e. those whose activation was increased for post-classical concepts).

The factor clusters are encircled and numbered for ease of reference in the text and their centroids are included in Table 2. These locations correspond to the four classes of processes evoked by the post-classical concepts.

Fig. 5. Factor scores of concepts at the extremes of the four factors.

A concept may have a high factor score for more than one factor; for example, potential energy appears as measurable, mathematical, and on the classical end of the post-classical dimension. In contrast, multiverse appears as non-measurable, non-periodic, and post-classical.

Fig. 6. Factor locations for the five factors (35 voxel clusters) are depicted on a rendered brain.

Colors differentiate the factors and greater color transparency indicates greater depth. Sample concepts from the two ends of the dimensions are listed. The post-classical factor locations include those whose activations were high for post-classical concepts (their locations are shown in Fig. 4) as well as those locations whose activations were high for classical concepts.