Hopkins et al. (2001) deserve much credit for having raised the bar for laterality studies in nonhuman primates by routinely studying such a large number of individuals and by developing well-defined tasks, such as the TUBE task, to try to quantify individual laterality in a reliable way. They also deserve credit for repeating prior studies to assess the consistency of results, because true replications are so rarely conducted (Palmer, 2000). Nonetheless, the reply by Hopkins and Cantalupo raises several important issues regarding their and others’ work on chimpanzee handedness. First, Hopkins and Cantalupo are correct that the unexpected patterns revealed by funnel graphs of the 1994 handedness data (Hopkins, 1994; Palmer, 2002) only became apparent because sample size (number of handedness observations per individual) varied. They are also correct that standardizing the number of observations per individual totally eliminates any possible sample-size effects: a funnel graph is only useful after the fact, as a form of quality-check of the data. The real question, however, is not as Hopkins and Cantalupo suggest, “How many observations to take per individual chimpanzee?,” but rather, “What aspects of the behavioral sampling protocol gave rise to the unexpected patterns in the first place?” Eliminating variation in the number of observations per individual merely eliminates the symptoms, but not the underlying cause(s), of the unusual distributions reported in Palmer (2002): why was right-handedness of individual chimpanzees more pronounced among those individuals for which fewer observations were recorded? If the underlying causes were not eliminated in the present study as indicated by Hopkins and Cantalupo, then they may still contribute to an apparent population-level right-handedness. Some further reflections by Hopkins et al. on the possible underlying causes of the odd patterns in the 1994 data might provide valuable insights that would help improve the design of future primate handedness studies. Second, my critique (Palmer, 2002) was also intended to illustrate how graphical presentation of data in as unreduced a form as possible (e.g., scatterplots, frequency distributions) is preferable to statistical summaries. Tabulated statistical summaries, although economical in terms of journal space, invariably obscure details about the data that may affect confidence in the results or that might suggest alternative interpretations. If, as was done in the original 1994 study, Hopkins and Cantalupo were to publish the raw data behind the statistical summaries tabulated in their reply and also in the more extensive study of Hopkins et al. (2001), this would allow the evidence for population-level right-handedness (at least among chimpanzees at the Yerkes Primate Research Center (YRPRC)) to be judged more fully and fairly. For example, Hopkins kindly provided to me raw data on the TUBE task for 109 chimpanzees at the YRPRC (from Hopkins et al., 2001), where the overall percent right-hand use was very similar to that reported in Hopkins (1994) for bimanual feeding (Table 1). Seventy-five of these individuals were also included in the 1994 study, and of these, handedness was based on more than 25 observations for 56 of them (for the justification to exclude individuals with 25 or fewer observations, see Palmer, 2002). The correlation between percent right-hand use during bimanual feeding (holding food in one hand while removing portions with the other) and percent right-hand use by the same individual in the TUBE task (holding a tube in one hand and scooping out peanut butter with a finger of the other) is not overly compelling (Fig. 1). Although these tasks are somewhat different, I am surprised to see so little consistency between tasks that require the simultaneous use of both hands in such a similar, coordinated fashion during feeding. If chimpanzees at the YRPRC exhibit population-level right-handedness for bimanual tasks, as the data in Table 1 suggest, rather little of this right-handedness appears to derive from consistent hand preferences by individual chimpanzees. Finally, as someone outside the field of primate laterality, I have always been troubled by handedness data obtained from captive animals, as have other primatologists (McGrew and Marchant, 1997). No matter how much care is taken to avoid introducing handedness biases in the sampling protocol
Read full abstract
Feb 1, 2011
Johanna Bruckner + 3
Open Access