“…In light of these empirical investigations, it appears that the best that a code model for the sounds of laughter can do here is to offer a statistical association—a natural code—between only two pairs: First, the acoustic cues making up what I will call Duchenne laughter is statistically paired with positive affective states and, second, non‐Duchenne laughter is paired with a heterogenous set of affective states (see Table 1). In other words, if we use ‘naturally means’ in a probabilistic sense (Bonard, 2023; Stegmann, 2015), then, as far as we know, Duchenne laughter naturally means positive affective states, non‐Duchenne laughter naturally means a heterogenous set of affective states, and there are no other acoustic categories of laughter that would naturally mean types of affective states. Let me underline that the empirical literature thus seems to show that we cannot acoustically distinguish between, say, nervous laughter, sardonic laughter, embarrassed laughter, and so forth (in Section 4, I will discuss attempts at devising a more sophisticated code using non‐acoustic cues).…”