Figure 6.

Mathematical behavioral results. Contrary to our previous linguistic paradigm, we found that subjects performed significantly worse on two-element combinatorial trials (the two-numeral addition condition in this case). Responses were both significantly slower (A) and less accurate (B) in this condition compared to the corresponding one-numeral control. We observed the same pattern of results in the List task, though to a lesser degree. ns, Non-significant; *p < 0.05; **p < 0.01; ***p < 0.001.