Reply to Vonk: Disentangling emotional contagion from its underlying causes

We thank Vonk (1) for her interest in our paper (2) in PNAS. We appreciate her concerns; however, several comments in her Letter are already discussed and supported by data in our paper. We thus respectfully disagree with her claims about the limitations of our study and theoretical interpretation.

Vonk argues that we should distinguish between “responding to others’ internal states rather than to behaviors” before claiming emotional contagion in animals. While we appreciate the importance of disentangling these mechanisms, we disagree that “to experience emotional contagion implies that observers perceive others’ emotions.” Emotional contagion refers to the transfer of emotional states, which does not necessarily imply representation of, or concern for, the other’s emotional state. We have repeatedly argued for a clear distinction on this matter (3), and thus in our paper we refrain from speculating about the mechanism facilitating emotional contagion. Moreover, Vonk suggests that our contagion findings reflect “expectation of reward” rather than emotion. Yet sensitivity to reward forms the foundation of the bias hypothesis: The expectation of reward is influenced by the emotional state, and so the observed evaluation of ambiguous stimuli reflects the underlying emotional valence (4, 5).

In addition, Vonk’s description of our experimental methodology and results seems misleading and might (partially) be based on an incomplete reading of our paper. While Vonk states that “results of cognitive bias tests are open to interpretation,” multiple studies have confirmed a priori predictions of environmental effects on cognitive bias (5). Indeed, the purpose of the bias test is to facilitate identification of “phenomena…not easily detectable by…behavioral measures” (2). Nonetheless, as we repeatedly state in our paper, we agree that bias tests are “best used in conjunction with other measures of emotional state.” While it is correct that the ratio of ambiguous to other cues is relatively high, the total amount of ambiguous cues is similar to past research and was selected to avoid false-positive findings (6). Although our subjects continued to respond to negative cues, we also disagree that “they did not fully learn when they would not receive a reward.” This interpretation would require dismissing the evidence that animals often exhibit difficulty inhibiting their responses in go/no-go tests (7). We therefore analyzed response latencies, to provide a more nuanced measure of reward expectation. Thus, our interpretation of ambiguous responses is based upon more than the fact that “birds continued to respond to [ambiguous cues].” Similarly, while the experience of the observer may influence their vicarious reaction (7), we show in the paper’s supplement that prior exposure had no effect on the observer’s bias test. Finally, although Vonk argues that the demonstrator bias data “did not support the manipulation,” we describe as well why interpretation of these responses was inhibited by unforeseen confounds, and that future research needs to overcome these limitations.

To conclude, we appreciate Vonk’s claim that our study presents a promising methodology, but we think she overstates its limitations.