How do you know whether to trust what someone is telling you? There’s ongoing debate about which cues are reliable, and how good we are at recognising deception. But now a new paper in Nature Communications reveals that we reliably take a particular pattern of speech pitch, loudness and duration as indicating either that the person lying or that they’re unsure of what they’re saying — and that we do it without even being aware of what we’re tuning into.
In an initial study, Louise Goupil at Sorbonne University, France, and her team manipulated the pitch, loudness and duration of a series of spoken pseudo-words (which sounded like they could be real words in French, but were not). Twenty native French speakers then listened to these words and rated the speaker’s honesty and also their certainty (honesty and certainty were investigated in two separate trials with the same participants, held one week apart). The participants were also asked how confident they were in their judgements.
The team’s analysis revealed that a single “prosodic signature” — that is, the same pattern of volume, pitch, and speed — was associated with perceptions of both honesty and certainty. Loudness (especially at the beginning of the word), a lower pitch towards the end of the word, a less variable pitch overall, and a faster pace of speaking were all associated with more honesty/confidence. The opposite patterns were associated with less of either. The team also found that the participants were more confident when making judgements about the speaker’s certainty.
Of course, uncertainty and deception are not the same thing. A subsequent study involving two groups of 20 participants revealed that context — in this case, background information about what the speakers were purportedly doing when they uttered the pseudo-words — allows us to use the same vocal signature to make judgements about either honesty or confidence, depending on the situation. However, while there was widespread agreement when it came to judging confidence, the group was more split when judging honesty. While the data showed that the prosodic signature informed all their judgements, some of the participants decided that the speaker was faking an honest voice. This suggests that we rely heavily on sensory evidence in inferring a speaker’s level of confidence, but our judgements about whether they’re being truthful or not are more complex.
Further questioning of this same group of participants revealed that though they did indeed rely on the three-factor prosodic signature in making their judgements, they were to a large extent unaware of just what they were tuning into. The team also studied additional participants whose native languages included English, Spanish, German, Marathi, Japanese and Mandarin Chinese — and found the same results. “Overall, these results demonstrate the language independence of a core prosodic signature that underlies both judgements of certainty and honesty,” the team writes.
The researchers suggest that the prosodic signature is tied to signs of cognitive effort: someone who has to make more of an effort with what they’re saying — either because they aren’t very sure or because they’re lying — will take longer to say it, for example, and use less emphasis. That could explain why it is apparently not culture-dependent, but fundamental to all people.
A follow-up study by the team did reveal that when participants heard words spoken with the prosodic signature of unreliability/dishonesty, these words “popped out” against effortless speech, grabbing the participants’ attention. So perhaps, starting in young childhood, we learn to spot these signs of cognitive effort, and learn to interpret them as indicating uncertainty or dishonesty. Alternatively, it might be part of a more ancient, innate system, shaped through evolutionary pressures to know whom to trust.
The team also found some gender differences in explicit judgements about whether people were lying/uncertain. For example, women were more likely to interpret certain/honest signatures as being faked. However, the data can’t reveal whether these differences were down to gender per se, or other related factors, such as anxiety or empathy.
Clearly, there are questions still to be answered. But it’s fascinating and, as the researchers write: “Our results add to the growing body of evidence suggesting that, contrary to decades of research arguing that humans are highly gullible, dedicated mechanisms actually allow us to detect unreliability in our social partners effectively.”
There are some immediate practical implications, too. We evolved, of course, to interact face to face — and we can only use this prosodic signature when someone is talking to us, not if they are giving us information via a keyboard and screen. Whatever the benefits of this increasingly common way of interacting, the silencing of this particular method for spotting liars and hustlers is clearly a cost.