We agree with Bard and colleagues’ general aim to highlight potential biases that may have influenced foundational developmental theories in human psychology (they note in particular two theories: Shared Intentionality Theory and Attachment Theory). As these hypotheses were founded on data derived primarily from WEIRD (Western, Educated, Industrial, Rich, and Democratic) populations, they may be prone to the same issues mentioned by Henrich and colleagues in their seminal work (2010). Thus, it is possible that these hypotheses will require reassessment based on new, more representative, data. However, the claim that developmental psychology is biased needs to be subjected to targeted testing in order to assess the existence of a bias, and if it is validated, quantify the extent and direction of it. Similarly, our current research (the Ape Research Index (ARI) project) involves identifying biases in comparative cognition research, in particular in where and how populations of captive chimpanzees are being tested (Bandini et al., Reference Bandini, Girish, Barnes, Boot, Tennie and Forss2025). Through an extensive literature review of all comparative cognition studies conducted with captive chimpanzees between 1950 and February 2024, we found that there is a significant difference in the extent and type of research history individual chimpanzees are collecting over their lifetimes. Whilst some chimpanzees are being tested on an almost daily basis for a large part of their lives at research-intensive facilities, others have only participated in one or two experiments so far. We hypothesise that differences in exposure to research will have an effect on performance in subsequent experiments and on the generalisability of findings deriving from captive chimpanzees that have reached a certain level of exposure to research. Therefore, it is possible that chimpanzees with much larger research histories are less representative of the capabilities of the species as a whole (potentially making them the WEIRD-counterparts within non-human ape populations; see also Leavens, 2010; Webster & Rutz, Reference Webster and Rutz2020; Bandini et al., Reference Bandini, Girish, Barnes, Boot, Tennie and Forss2025). Bard and colleagues acknowledge the existence of this potential bias as well (e.g., “This systematic weakness permeates the contemporary scientific literature, and results in large part from overgeneralizing from both the human samples…and the ape samples (taking captive apes that are convenient to sample as representative of their respective species)” (sect. 2.5, para. 2)).

Whilst we agree with the idea that data should, ideally, be collected “worldwide,” “in situ” with “local” and “diverse” populations, there are additional important considerations that should be made when testing these populations. With these aspects in mind, we propose an extension to Newson et al., Reference Newson, Buhrmester, Xygalatas and Whitehouse2021 WILD acronym and Bard and colleagues’ five steps.

The first factor we propose to be included is “Experience in research” (E). The effect of experience on the developmental trajectory of specific traits and cognitive performance (in both a positive and negative direction) has been closely studied in humans (e.g., Bouchard et al., Reference Bouchard and McGue1990; Nelson et al., Reference Nelson, Fox and Zeanah2023). Whilst most studies have focused on the effect of experience accumulated over longer timeframes, recent experiences may also result in shorter-term effects (Brooks et al., Reference Brooks, Yoshimura and Taki2021). In the context of research paradigms, it is likely that participants will learn about specific strategies and associated outcomes as they continue to take part in cognitive research, and this experience may transfer across cognitive domains, as well. To ensure the generalisability and robusticity of developmental theories, it is essential to test non-WEIRD populations (as argued by the original authors). However, due to the often complex logistical and ethical considerations required when testing some non-WEIRD populations (including those that the authors refer to as: “traditional subsistence farmers, and small-scale foraging people”), it is possible that multiple researchers converge on the same groups (or individuals) that are more easily accessible, or are more open to testing of this nature. Repeated exposure to testing may then have an effect on performance and the type of data collected, on both an individual and group level. Thus, we suggest that the extent and type of existing experience in research be accounted for on an individual level when testing the generalisability of developmental theories. On a larger scale, future studies should test the magnitude and direction of the effect of research exposure on performance in human participants. This will ensure that future studies generate more robust and replicable data that are less vulnerable to the biases highlighted by the authors.

The second factor we suggest should be included is “Replication of experiments” (R). Whilst the original authors highlight the need to replicate findings from different populations (a sentiment that is largely uncontroversial), we argue that it is equally important to replicate experiments with sub-groups within a population, whenever possible. As discussed in the target article and recent extensions of the WEIRD literature, not all WEIRD or non-WEIRD populations are the same, or will behave similarly in experimental paradigms (Chaudhary et al., Reference Chaudhary, Misra, Bansal, Valsiner and Singh2022). Furthermore, as mentioned above, existing experience may have a lasting effect on behaviour. Thus, one way to counteract this potential confounding factor is to include various distinct sub-groups in each study. As Bard and colleagues also highlight, no single group can be representative of a species as a whole. The same may be true for sub-groups, where limiting data collection to only one group (or sub-group), and without accounting for individual-level experience histories, may result in both Type I and Type II errors.

In our research, we found that only 15% of comparative cognitive studies with captive chimpanzees included more than one study sample per experiment (Bandini et al., Reference Bandini, Girish, Barnes, Boot, Tennie and Forss2025). Ideally, multiple groups should be tested per research question (a power analysis would be required for exact counts). This should allow for the identification of underlying group differences and patterns, thus better accounting for inter- and intra-group variation. Taking into account experience in the sample selection and/or analysis, and replicating experiments with multiple groups will generate more robust insights, increasing the validity of future studies in assessing the universality of specific phenomena. As such, we propose that the WILD framework be extended to include “Experience” and “Replication.” After all, why stop at WILD, when you could go WILD-ER?