Adolescents' Neural Processing of Risky Decisions: Effects of Sex and Behavioral Disinhibition

Abstract

Background: Accidental injury and homicide, relatively common among adolescents, often follow risky behaviors; those are done more by boys and by adolescents with greater behavioral disinhibition (BD).

Hypothesis: Neural processing during adolescents' risky decision-making will differ in youths with greater BD severity, and in males vs. females, both before cautious behaviors and before risky behaviors.

Methodology/principal findings: 81 adolescents (PATIENTS with substance and conduct problems, and comparison youths (Comparisons)), assessed in a 2 x 2 design (

Patients: Comparisons x Male:Female) repeatedly decided between doing a cautious behavior that earned 1 cent, or a risky one that either won 5 or lost 10 cents. Odds of winning after risky responses gradually decreased. Functional magnetic resonance imaging captured brain activity during 4-sec deliberation periods preceding responses. Most neural activation appeared in known decision-making structures. PATIENTS, who had more severe BD scores and clinical problems than Comparisons, also had extensive neural hypoactivity. Comparisons' greater activation before cautious responses included frontal pole, medial prefrontal cortex, striatum, and other regions; and before risky responses, insula, temporal, and parietal regions. Males made more risky and fewer cautious responses than females, but before cautious responses males activated numerous regions more than females. Before risky behaviors female-greater activation was more posterior, and male-greater more anterior.

Conclusions/significance: Neural processing differences during risky-cautious decision-making may underlie group differences in adolescents' substance-related and antisocial risk-taking. Patients reported harmful real-life decisions and showed extensive neural hypoactivity during risky-or-cautious decision-making. Males made more risky responses than females; apparently biased toward risky decisions, males (compared with females) utilized many more neural resources to make and maintain cautious decisions, indicating an important risk-related brain sexual dimorphism. The results suggest new possibilities for prevention and management of excessive, dangerous adolescent risk-taking.

Fig 1. Game; Participants' Behavioral Disinhibition Scores.

(a) Colorado Balloon Game. A: Decision Trial begins, yellow illuminated 4 sec. B: Chosen press executed during 0.5-second green light. Next, red light signals outcome, 3.5 sec. If left press in B, then C (counter increases 1 cent, dull sound, no change in balloon). If right press in B, then either D (counter decreases 10 cents, frowney face, popping sound, balloon shrinks), or E (counter increases 5 cents, smiley face, coin-drop sound, balloon puffs up). F: Fixation screen 2–4 sec. G: Directed Trial begins with 4-sec yellow half-light. Illuminated side (left here) indicates side to press when light turns green. Counter: current earnings. Then, the B-F sequence from the paired Decision Trial follows exactly, but participant knows that "the computer is playing the game now", and that only the counter increase of 2 cents upon the directed press affects participant. (b) 81 Participants’ BD Scores. Crosses: means, standard deviations.

Fig 2. Decline in risky presses as probability of loss increased.

Four groups: patient males, comparison males, patient females, comparison females. Raw (not fitted) data shown. Comparison males' error bars: red to reduce confusing overlaps. Trials required a choice: a cautious response earned 1 cent; a risky response either won 5 or lost 10 cents. Probability of winning after risky choices declined from 0.78 to 0.22 as game progressed.

Fig 3. Cortical regions near scalp where groups' activations differed significantly.

These regions are potentially accessible to transcranial direct current stimulation. Row labels: Cau, before cautious responses. Ris, before risky responses. Cmp, comparison participants. F, females. M, males. Pt, patients. Three rows of possible activations not shown: (a) Cau: Pt>Cmp and (b) Ris: Pt>Cmp, were both devoid of activation; (c) Cau: F>M had activation only at site marked by Green pointer tip in Ris: M>F. Significant t-values range from 2.64 (dark red) to 7.76 (white); details in Tables 4 and 5.

Fig 4. Three brain regions showing significant male:female x comparison:patient group interactions (cf., Tables 4 and 5).

Dashed lines, males. Solid lines, females. Cmp, comparison participants. Pt, patients. Activation intensity is the mean activation difference (Decision Trial – Directed Trial) for all voxels in the cluster, so negative values indicate relative deactivation in Decision Trials, compared to Directed Trials. Standard errors of the means for each of these points lie in the narrow range ± 0.064–0.094.

Fig 5. Assessing possible confounds by changing samples or procedures.

Clusters activating significantly differently in groups appear as shadows on a 2-dimensional surface; activations preceding cautious or risky behaviors are shown separately. Abbreviations: Cmp, comparison participants. F, females. M, males. Pt, patients. Row A: entire sample, data analyzed as in Tables 4 and 5. Row B: same sample, additional adjustment for depression severity (Carroll Rating Scale). Rows C-E: analyses as in Row A, but excluding: left-handed participants (Row C: 2 patient males, 1 comparison male, 3 patient females, no comparison females); or participants receiving psychotropic medications (Row D: see Table 1); or recent regular tobacco smokers (Row E: 6 patient males, 1 comparison male, 20 patient females, 1 comparison female). All images modified identically to increase contrast between the white brain figure and its gray background.