The Question

Mindreading is characterised by confident local agreement and variation between groups (and, as we will see later, experts). What explains this combination?

Confident Local Agreement

Development

There is evidence for stable patterns of development within a country and variation between countries.

Wellman & Liu (2004) developed a scale of mindreading tasks involving desire, emotion, knowledge and belief. This has been widely used to study the order in which children come to succeed at different aspects of mindreading.^[1]

A meta-analysis by Yu & Wellman (2024) covers how children from 22 countries progressively improve on different mindreading tasks involving desire, emotion, knowledge and belief. They find variation between countries in the order of diverse-belief and knowledge-access understanding.^[2]

Variation on Agency Attributions

Ethnographers describe a range of contemporary views concerning individual and shared agency. Among the Northern Ojibwa, certain stones are regarded as acting voluntarily and may even talk (Hallowell, 2019), and some guides in the Alps regard mountains as living communicators who send signals the guides learn to understand (Chamel, 2023). Improvising musicians in the Netherlands report that music gains a life of its own, and becomes something that provides direction and requires service.^[3] Of the things that to me seem to be inanimate objects and nothing more, some are regarded by donso hunters in Burkina Faso as agents. These jo can act in the world. If correct procedures are followed and appropriate moral conditions met, they may respond to human actions (Ferrarini, 2025, p. 235). Indeed, Ferrarini reports not merely using a jo as a tool but cooperating with it (Ferrarini, 2025, p. 235), and interprets sacrifice as sharing the results of a successful hunt (Ferrarini, 2025, p. 239). Or, to take another example, certain plants appear to Makushi people in Guyana as helpers to be called on. When approached in the right way, these special plants can become voluntary partners in actions such as diagnosis and healing (Whitaker, Tamboli, Daly, & Lewy, 2024, p. 353). And for a Pemon kanaima in Venezuela, a plant may be regarded as the dominant partner in a relationship (Whitaker et al., 2024, p. 355).

Animism is not, or not always, associated with generally more inclusive views on what counts as an agent. ojalehto, Medin, & García (2017, p. Experiment 2) investigated how US and Ngöbe participants attribute agency to plants, abiotic kinds (non-living, natural things) and complex artefacts. They found that US participants were comparatively more likely to attribute agency to computers and other complex artefacts, whereas Ngöbe participants were more likely to attribute agency to plants and abiotic kinds.Neither group seems to have a straightforwardly more inclusive notion of agency than the other—and nor is it entirely clear which, if either, is animist.^[4]

Animist views tend to be quite discriminating, and different groups discriminate differently. This applies to views on which things are agents as well as on which events are exercises of agency. To illustrate, the Northern Ojibwa are far from regarding any stone as an agent. On one occasion, a stone followed the path of a man but ‘[t]he animate behavior of a stone under these circumstances [...] was not a voluntary act initiated by the stone’ (Hallowell, 2019, p. 25). As this suggests, animist views do not appear to involve a less principled or less systematic division between exercises of agency and things that merely happen.

Even where people are in broad agreement on facts about behaviours, they might disagree about which attributions those behaviours warrant. ojalehto, Medin, & García (2017a) presented their participants with drawings of plants and other entities plus information about each thing’s needs. Ngöbe and US participants alike expected the unfamiliar plants to move towards things that are beneficial (Experiment 1A). There was also some agreement in that both groups attributed some feelings and mostly refrained from attributing seeing to plants. But the two groups diverged on whether the unfamiliar plants could want things or feel pain (Experiment 1B). Apparently, then, agreement on relevant behaviours need not settle which agential characteristics things have.^[5]

Variation on Mental State Attributions

Strange Stories in Australia and Chile

People seem to do slightly better at mindreading in culturally familiar situations.

Perez-Zapata, Slaughter, & Henry (2016) gave Australian and Chilean adults two sets of ‘strange stories’ mindreading tasks (Happé, 1994; White, Hill, Happé, & Frith, 2009). One set was adapted to Chilean cultural context (names, places, pictures were changed) while keeping the mindreading demands as similiar as possible. They found an interaction: each group of participants did slightly better on the strange stories adapted to their cultural group.

Relatedly, Adams et al. (2010) found an advantage for culturally familiar stimuli in attributing mental states on the basis of photographs of eyes in a study of Japanese and white American participants. (‘Reading the Mind in the Eyes’; Baron-Cohen, Wheelwright, Hill, Raste, & Plumb, 2001).

Example Strange Story

source: White et al. (2009, pp. 1109–1110)^[6]

‘During the war, the Red army captures a member of the Blue army. They want him to tell them where his army’s tanks are; they know they are either by the sea or in the mountains. They know that the prisoner will not want to tell them, he will want to save his army, and so he will certainly lie to them. The prisoner is very brave and very clever, he will not let them find his tanks. The tanks are really in the mountains. Now when the other side asks him where his tanks are, he says, ‘‘They are in the mountains.’’

Q: Why did the prisoner say that?

2 points—reference to fact that other army will not believe and hence look in other place, reference to prisoner’s realization that that’s what they’ll do, or reference to double bluff

1 point—reference to outcome (to save his army’s tanks) or to mislead them

0 points—reference to motivation that misses the point of double bluff (he was scared)

False Belief in Vanuatu and Samoa

In Vanuatu, one study found variation in the age at which groups first pass standard false belief tasks. The range was around 7 years of age for urban children, 9 years for rural Tanna children, and at none of the tested ages did a majority of the island (Nguana) cohort pass (Dixson, Komugabe-Dixson, Dixson, & Low, 2018).

This illustrates differences among Vanuatu communities, and between Vanuatu and Western communities (where children typically pass the false belief tasks much earlier).^[7]

(Dixson et al., 2018; Mayer & Träuble, 2012). These findings make it difficult to assume that the familiar Western developmental curve is simply the universal timetable revealed with more or less noise.

But Callaghan et al. (2005)

Mindreading and Nicaraguan Sign Language

In studies of profoundly deaf adults who were brought up by non-signing parents and invented or learned what became Nicaraguan Sign Language (NSL) through school, Pyers & Senghas (2009) compare two cohorts of signers. The first cohort had been exposed to an early form of NSL; the second, younger cohort encountered a more developed form of NSL. The two cohorts attended the same school, had teachers in common, and were socio-economically similar.

As adults, the two cohorts show very different performance on mindreading tasks, including false belief and knowledge tasks Pyers & Senghas (2009).

First-person Authority

Mitchell, Teucher, Kikuno, & Bennett (2010) asked children in Britain and Japan how much they knew about their own feelings and mental states (feeling happy or thinking, for example) and how much a parent or teacher know. They find that Japanese children overall attribute less first-person authority than British children (adults know relatively more in their estimation).^[8]

Was that Intentional?

Judgements about which side-effects of an agent’s action are intentional vary across cultures.

Robbins, Shepard, & Rochat (2017) find that in two groups, from Vanuatu and Samoa, participants are morel likely to judge helpful side-effects to be intentional than they are to judge harmful side-effects to be intentional. This reverses the well-studied ‘Knobe effect’ (Knobe, 2003) they found in other participant groups. Robbins et al. (2017, p. Experiments 2–4) also find that attributions of intention are linked to the agent’s status in participants from Vanuatu and from Samoa, but not from the US.

Clark, Bauman, Kamble, & Knowles (2017) find Indian participants more likely than US participants to attribute intention to an agent concerning helpful side-effects of their actions (with no difference for harmful side-effects).

Do Your Intentions Matter?

McNamara, Willard, Norenzayan, & Henrich (2019) investigate the relative importance of negative intentions and negative outcomes. How much does an intention to poison you matter, and how much does it matter whether the poisoning occurs or not? McNamara et al. found that Yaswan Fijians weigh negative outcome more, and negative intent less, compared to the Indo-Fijian and US/Canadian groups.

Variation in Experts

Starmans & Friedman (2020) studied attributions of knowledge in philosophers, academics from other disciplines and non-academics. They found that philosophers respond differently to Gettier cases (where they are attribute knowledge less relative to their judgements about other scenarios), and to scenarios involving sceptical pressure (where they attribute relatively more knowledge).