Understanding how much to morally value animals of different species relative to each other and relative to humans is highly consequential for EA.

Understanding how laypeople actually do morally value animals in practice is consequential for EA in a variety of ways:

• We may give some weight to moral judgments of large numbers of individuals, with a broader set of moral views than the EA community, on the grounds of moral uncertainty or peer disagreement
• We may think that identifying influences on moral judgement can highlight potential biases in our own judgments, and potentially offer debunking arguments against our own intuitions
• Knowing individuals' views may be of practical relevance to our actions e.g. showing which animals the general public are highly sympathetic towards and which they are not, to suggest which campaigns would or would not likely be successful
• Such studies can also highlight predictors of individuals caring more or less about non-human animals, which can be valuable for similar reasons.

This recent study by Miralles et al (2019), elicited judgments from 3500 raters (2347 after exclusions) about their relative levels of empathy or compassion for animals of different species. Empathy was measured by presenting individuals with the statement "I feel like I’m better able to understand the feelings or the emotions of..." and a choice of two pictures of animals, and compassion was measured with "If these two individuals were in danger of death, I will spare the life of [choice among a pair of pictures] as a priority."

Both empathy and compassion scores decreased sharply with increasing phylogenetic distance from humans. However, beyond a certain point scores on both measures stopped decreasing.

The results also highlighted animals for which respondents had relatively high compassion compared to empathy or vice versa, although these were fairly highly correlated. For example, respondents had relatively high compassion toward (or willingness to save) a platypus relative to their levels of empathy for the platypus and relative to other animals for which they had similar levels of empathy (such as lizards or alligators). Empirical data of this kind seems to have clear advantages over simply relying on our intuitions about which animals individuals will care about.

Limitations

The study merely offered participants with a forced choice between two different animals, and so the analyses are based around the probability of animals being chosen in such a forced choice. This cannot tell us directly about the extent of compassion felt for different animals, which might vary by several orders of magnitude. Studies posing explicit tradeoffs (e.g. how many animals of a certain species are worth the same morally as one animal of another species (see Alexander (a) (b)), or asking willingness to pay to save animals of different species might be more powerful for this purpose.

The study compared empathy and compassion scores to evolutionary divergence time, but there are a whole host of other factors varying across species which are potentially relevant and may confound these judgments). The study posits that "the phylogenetically closer a species is to us, the more it shares common traits with us. Our results could be explained by the fact that many of these traits may arouse sensory biases." However, other traits, such as neuron count (see Alexander citations above) likely also correlate with phylogenetic distance and with the moral value ascribed to animals (although Rethink Priorities will be publishing further work soon describing how this may be misleading). Further work would be necessary to investigate how these judgments correlate with features such as neuron count and perceived similarity to humans. Future work either using the paradigm employed in this study or the paradigms employed in Alexander's and Rethink Priorities' studies or something similar could also employ a wider variety of species and stimuli to explore the influences of different features on perceived moral value.

A significant potential benefit of this study was that it employed photos of animals of different species, rather than simply asking in abstracta about comparisons of animals of different species. Using visual prompts may produce more ecologically valid results and may make visible traits (such as similarity to humans) more salient (which may be a positive or a negative effect overall). Using photos of specific animals from different species also raises tricky issues concerning stimulus sampling. Though the authors used 4 photos per species, it is not clear to me that this is an adequately representative sample or that they adequately accounted for the potential effect of different photos in their analysis.

Conclusion

Overall I think many EAs would benefit from this paper and that more research in this area would likely be of value to the EA community.