Fig. 4.

Energy maximization does not explain nutrient preferences. (A) Schematic of cumulative choice trajectory between isocaloric high-sugar (LFHS) and high-fat (HFLS) rewards in reward space. (Inset) Proportions of fat and sugar to matched energy content in the two rewards. (B) Cumulative choices between isocaloric high-sugar (LFHS) and high-fat (HFLS) rewards for the three animals (black: mean trajectory of actual choices, gray: single-session trajectories; colors: simulated choice trajectories based on reference strategies maximizing calories, fat, or sugar). All three monkeys’ choice trajectories were biased toward high-sugar reward. (C) Schematic of cumulative choices in nutrient space polar coordinates, showing dynamic fat–sugar trade-off (slope: relative fat–sugar intake ratio and radius: trial progression). (D) Choice trajectories transformed from reward space into nutrient space (same sessions as in C; black: actual choices; colors: reference simulated choices). (E) Model comparison based on Akaike Information Criteria (AIC) favored the nutrient model with separate fat and sugar regressors over the energy model in explaining reward choices. (F) Higher sensitivities to sugar compared to fat despite matched energy content suggested by differences in logistic-regression coefficients for isocaloric sugar and fat levels (P < 0.001, Wilcoxon signed-rank test).