Cross-cultural comparison of subjective concepts requires a common denominator serving as a basis of comparison. For instance color concepts have been compared across cultures by asking informants to divide the optical spectrum into named colors. Unfortunately similar elegant physical parameters are not always available. A case in point may be the cross-cultural comparison of conceptions of causality discussed recently in Frontiers by Widlok (2014). A current approach has contrasted “Western” concepts with non-Western “religious” and “magical” concepts. Widlok has questioned this approach and, relying on an exploration of relevant ethnographic observations, he has proposed an alternative approach of causal cognition involving two parameters: (a) the temporal dimension of sequence, and (b) the concept of agency. One can very well imagine “time” as a basis of cross-cultural comparison as the optical spectrum is. However, “agency” seems quite a different matter. Exploring concepts of agency in the ethnographic literature, Widlok comes upon distinctions such as between “natural” causes versus causes involving “personhood.” However, it may not always be clear how similar distinctions may map particular culture-specific conceptions of causality involving, for instance, magical and spiritual forces. Hence in the following a formal non-physical basis of comparison is proposed. In order to compare cognition across cultures we should proceed from a universal feature of cognition. A similar feature would be the organization of cognition into entities and relations reflected by the linguistic noun-verb distinction (Bever, 1970). Using entities and relations as basic cognitive units of analysis, cross-cultural research may tie in with studies involving predominantly Western educated subjects reviewed in the theoretical and discussion sections of two articles available by internet (Peeters, 2004; Peeters and Hendrickx, 2007). For instance, if relations are represented as vectors (arrows), more informational weight seems to be attached to origins (arrow-tails) than to terminals (arrow-heads). Being informed that John likes Pat, subjects attribute likableness to John rather than to Pat. In order to be found likable, Pat should like John in turn (Peeters, 1983). This is in line with Widlok's (2014) observation that Westerners locate causality in the agent rather than in the object acted upon. Replicating the related studies with non-Western Dinka might yield opposite results likableness being attributed to Pat rather than to John. An important feature of relations concerns reflexivity. A relation is “reflexive” when it forms a loop connecting an entity with itself, and it is “non-reflexive” when it connects one entity with another entity. An all-important structural feature of the architecture of cognition seems to boil down to the double possibility to have reflexivity either attended to or ignored. Attending to reflexivity, subjects deal with entities as “self” and “other” (SO-thought), whereas ignoring reflexivity, they deal with them in the third-person, as “he,” “she,” and “it” (3P-thought). One general finding is that SO-thought is a hallmark of cognitive representations that are intuitively associated with personhood. 3P thought marks rather impersonal cognitive representations such as in hard natural sciences. For instance, given the information that John regales someone with cake, an observer may attribute generosity to John only if “someone” is not John himself but another person (SO-thought). However, in order to draw the rather impersonal inference that John has resources enabling to get cake, the observer may ignore whether the gourmand is John himself or someone else (3P-thought). The connection of attention for reflexivity with attributions of personhood may enable to detect hidden homunculi that, according to Widlok (2014), may be concealed in the curtains of mechanical causal explanations. Alternatively, the disregard of reflexivity may reveal quasi-mechanical thinking about the action of, for instance, magical powers that otherwise might be misconceived as intentional actions of quasi-human ghosts. Notice that there may be no one-to-one relationship between SO/3P thought and the use of correspondent linguistic codes in the communication of those thoughts. SO/3P thought is diagnosed indirectly from the ways subjects accomplish particular experimental tasks that are often more complex than the examples presented. It may require some inventiveness to have them adjusted to cross-cultural research that may proceed from hypotheses like the ones below.