Can impulsivity evolve in response to childhood environmental harshness?

Previous studies have suggested that human impulsivity is an adaptive response to childhood environmental harshness: individuals from families of low socioeconomic status (SES) tend to be more impulsive. However, no studies have tested the evolvability of this reaction norm. This study examined whether (a) impulsivity is associated with higher fitness among individuals from low SES families, while (b) it is associated with lower fitness among individuals from high SES families. We assessed three indices of impulsivity (temporal discounting, risk taking and fast/slow life history strategy), childhood SES and five proxy indices of fitness (number of children, lifelong singlehood, annual household income, subjective SES and life satisfaction) of 692 middle-aged participants (40-45 years old). None of the results supported the evolvability of the impulsivity reaction norm, although low childhood SES was associated with lower fitness on every proxy measure. Impulsivity (operationalised as the fast life history strategy) was associated with lower fitness regardless of childhood SES.

Note. Number of children was submitted to Poisson regression analyses; marriage experience was submitted to logistic regression analyses. Bold coefficients indicate significant difference in the hypothesised direction. Bold italic coefficients indicate significant difference opposite to the hypothesised direction.

Figure S1
Distribution of fitness indices of four groups divided into the upper/lower tertiles of childhood SES and impulsivity.
Note. The hypothesised inequality predicts a horizontally mirrored J-shape: highest, second-highest, lowest, and second-lowest for Groups 1, 2, 3, and 4, respectively. Results based on childhood SES × temporal discounting grouping are presented as Figures S1a to S1e. Results based on childhood SES × risk-taking grouping are presented as Figures S1f to S1j. Results based on childhood SES × Mini-K grouping are presented as Figures S1k to S1q. The dependent variables were the number of children for Figures S1a, S1f, and S1k, marriage experience for Figures S1b, S1g, and S1m, annual household income for Figures S1c, S1h, and S1n, subjective SES for Figures S1d, S1i, and S1p, and life satisfaction for Figures S1e, S1j and S1q.  1  2  3  4  Group  1  2  3  4  Group  1  2  3  4  Group  1  2  3  4  Group  1  2  3

Figure S2
Distribution of fitness indices of four groups median split into by childhood SES and impulsivity.
Only individuals whose annual household income decreased from the last year were included.
Note. The hypothesised inequality predicts a horizontally mirrored J-shape: highest, second-highest, lowest, and second-lowest for Groups 1, 2, 3, and 4, respectively.   Note. Number of children was submitted to Poisson regression analyses; marriage experience was submitted to logistic regression analyses. Bold coefficients indicate significant difference in the hypothesised direction. Bold italic coefficients indicate significant difference opposite to the hypothesised direction.

Figure S3
Distribution of fitness indices of four groups median split into by childhood SES and impulsivity (Male Participants).
Note. The hypothesised inequality predicts a horizontally mirrored J-shape: highest, second-highest, lowest, and second-lowest for Groups 1, 2, 3, and 4, respectively. Results based on childhood SES × temporal discounting grouping are presented as Figures

Figure S4
Distribution of fitness indices of four groups median split into by childhood SES and impulsivity (Female Participants).
Note. The hypothesised inequality predicts a horizontally mirrored J-shape: highest, second-highest,     Note. Childhood SES and impulsivity were standardised for the set of these multiple regression analyses. Number of children was submitted to Poisson regression analyses, and marriage experience was submitted to logistic regression analyses. For the remaining dependent variables (i.e., annual household income, subjective SES, and life satisfaction), reported regression coefficients are standardized coefficients. For the two shaded cells (one significant and one marginally significant interaction effects), we conducted a series of simple slope tests. Neither of the interactions was consistent with the hypothesis (the slope is positive for low childhood SES group and negative for high childhood SES group). .050 .803 Note. Childhood SES and impulsivity were standardised for the set of these multiple regression analyses. Number of children was submitted to Poisson regression analyses, and marriage experience was submitted to logistic regression analyses. For the remaining dependent variables (i.e., annual household income, subjective SES, and life satisfaction), reported regression coefficients are standardized coefficients. For the shaded cell (the significant interaction effect), we conducted a simple slope test. The interaction was not consistent with the hypothesis (the slope is positive for low childhood SES group and negative for high childhood SES group). Note. Childhood SES and impulsivity were standardised for the set of these multiple regression analyses. Number of children was submitted to Poisson regression analyses, and marriage experience was submitted to logistic regression analyses. For the remaining dependent variables (i.e., annual household income, subjective SES, and life satisfaction), reported regression coefficients are standardized coefficients. For the two shaded cells (one significant and one marginally significant interaction effects), we conducted a series of simple slope tests. Neither of the interactions was consistent with the hypothesis (the slope is positive for low childhood SES group and negative for high childhood SES group). Note. Childhood SES and impulsivity were standardised for the set of these multiple regression analyses. Number of children was submitted to Poisson regression analyses, and marriage experience was submitted to logistic regression analyses. For the remaining dependent variables (i.e., annual household income, subjective SES, and life satisfaction), reported regression coefficients are standardized coefficients. For the three shaded cells (one significant and two marginally significant interaction effects), we conducted a series of simple slope tests. Of the three, the marginally significant interaction between childhood SES and temporal discounting on number of children tended to be supportive of the life history hypothesis (the slope is positive and significant for low childhood SES group and negative but non-significant for high childhood SES group), while the other two interactions were not consistent with the hypothesis.