Recognizing someone you know is actually not a simple cognitive task – it requires you interpret the information you’re currently receiving through your senses, and then link back to a previously-formed conceptual representation you have of the individual in question. It’s especially difficult if you are acting cross-modally, for example matching someone’s voice to a photograph or vice versa.
Recently, two separate studies have shown that rhesus macaque monkeys (Macaca mulatta) are quite up to this challenge, reflecting that they possess a considerable degree of social memory and engage in complex conceptual thinking about other individuals.
In the first study1, published earlier this year in Proceedings of the National Academy of Sciences, a French research team headed by Julian Sliwa of the University of Lyon confirmed that rhesus macaques are able to spontaneously match the faces of known macaques and humans to their voices.
In their experiments, the research team gave six macaques a large number of tests in which they played short voice samples of known individuals (coos and grunts for other macaques, short French sentences and phrases for humans) and then measured how long the macaques spontaneously looked at cropped photographs of two known faces, only one of which matched the voice they had heard. The researchers statistically analyzed whether the macaques spent more time looking at specific photos after hearing the matching voice than they did after hearing a different voice, and found that the macaques did indeed stare significantly longer at a photo – whether of another macaque or a human – if the matching voice had been played first.
In reviewing individual performance, the researchers observed that five of six of the macaques displayed this effect overall, and that a greater number were better at recognizing photos that matched human voices than ones matching the voices of fellow macaques (the researchers noted that they were surprised at this finding, but pointed out that perhaps the explanation was that there were more useful auditory cues in the human speech samples than there were in the monkey coo vocalizations). Finally, the researchers found that five of the six monkeys showed preferences for specific faces, spending an especially long time looking at matching” photos of certain individuals – often a “neighbor” monkey or the researcher who was their main caregiver.
The researchers concluded that rhesus macaques can recognize individuals, linking together abbreviated visual and auditory perceptual cues (small, two-dimensional photos and short sound samples) to spontaneously identify other macaques and socially-relevant humans, and even to reflect the preference biases they have towards specific individuals.
At the Movies
The second study2, published last week in PLOS ONE, extended the findings to show that rhesus macaques can also recognize photos of other macaques whom they had seen during video clips, an additional challenge because specific features can be harder to identify in dynamic movies than in still images.
In this study, researchers led by Ikuma Adachi of the Yerkes National Primate Research Center began by training five macaques to watch brief silent video clips of familiar individuals before identifying which of five randomly placed photos represented the individual in the video. At first the macaques were allowed to continue to look at the last frame of the video before having to choose the correct photo, but in a second phase of the experiment the screen went black after the video was played, and the monkeys had to choose the correct photo after a time lag.
In each case the macaques became proficient at the task, even performing well after seeing videos taken from a novel perspective that was substantially different than the view in the training videos. Thus, their performance suggested that they were able to recognize specific features of known individuals as they appeared in dynamically-changing scenes in a range of videos, and then extract that information to identify those individuals later on in still images.
Next, the researchers repeated the testing, but this time they tweaked the conditions by playing a brief vocalization right after showing the last frame of the some of the videos – either a vocalization of the macaque in the video (the “congruent condition”) or of a different macaque (the “incongruent condition”). Only two of the macaques participated in this testing, as apparently the other three weren’t comfortable with working in the sound isolation booth necessary for this phase.
The researchers found that the macaques, who had never been trained to use vocalizations to guide their test responses, continued to be good at choosing the “correct” photo, but that when they made errors, they were statistically more likely than chance to pick the image of the vocalizing monkey, rather than the one in the video.
In other words, hearing the vocalizations systematically biased the macaques’ choice behavior, indicating that the voices may have activated visual representations of the vocalizing monkeys that occasionally superseded the impact of what had been seen in the video. Again, the macaques were demonstrating how they processed the information they used to recognize information cross-modally.
So, clearly “see no evil” is linked to “hear no evil” – perhaps we’ll see how “speak no evil” fits into the picture in a later post.
1Sliwa J, Duhamel JR, Pascalis O, & Wirth S (2011). Spontaneous voice-face identity matching by rhesus monkeys for familiar conspecifics and humans. Proceedings of the National Academy of Sciences of the United States of America, 108 (4), 1735-40 PMID: 21220340.
2Adachi, I., & Hampton, R. (2011). Rhesus Monkeys See Who They Hear: Spontaneous Cross-Modal Memory for Familiar Conspecifics PLoS ONE, 6 (8) DOI: 10.1371/journal.pone.0023345.