Friday, 11 November 2011

"Change deafness" - the scant attention we pay to the voice on the end of the phone

Our perception of the world is so restricted by the brain's finite attentional resources that large changes to the visual scene can occur without us noticing. Psychologists have studied this extensively and they call it "change blindness". But what about our limited vigilance to the world of sound? In a new study, Kimberly Fenn and her team have tested whether people notice when, mid phone-conversation, the person they're talking to changes. They found that unless there was a change of gender, most people didn't notice they were talking to someone else - a phenomenon the researchers call "change deafness".

Across five experiments, Fenn's team followed a similar procedure. Participants were interviewed on the telephone, ostensibly as part of a study into memories of smells. A young female interviewer greeted them, explained that there'd be twelve questions, then proceeded to fire away. After the third question, a different interviewer, usually another female, took over the questioning without warning or announcement (the four women who played the role of interviewer across the different experiments had voice frequencies of 200Hz, 202Hz, 218Hz, and 239Hz). After the twelfth question, participants were told the phone would be passed to a "supervisor". The supervisor took the phone, introduced herself, and asked progressively more specific questions to find out if participants had noticed the earlier voice change, ranging from "Did anything unusual happen during the interview?" to, "Did the experimenter's voice change at all during the interview?"

In the first two experiments, just 1 person out of 16 (6 per cent) and 1 out of 24 (4 per cent), respectively, noticed the voice change, even after they were asked about this directly. Moreover, none of the participants made any mention during the interview when the voice of the interviewer changed.

After the initial interview, but before the supervisor questions about a voice change, the participants were played recordings of the two interviewer voices and asked by the supervisor to say which was the voice of the interviewer (remember, at this point nearly all of them thought there was just one interviewer). Participants picked out the first interviewer voice just as often as the second voice - so it's not that one was particularly more memorable or dominant. However, presented with either one of the interviewer voices and a strange, unfamiliar voice, most participants (74 per cent) correctly picked out the interviewer voice. This means that in spite of the "change deafness" some aspects of the interviewer voices must have been encoded.

In another experiment, participants were warned in advance that the voice of the interviewer might change at some point during the interview. In this case, 75 per cent correctly reported afterwards that the voice of the interviewer had changed, and six of these nine participants knew the precise moment that the switch occurred. This suggests "change deafness" doesn't occur because we're incapable of detecting a change, but because in usual circumstances we don't bother paying enough attention to voices to detect such a change. This makes strategic sense, leaving more processing resources available for focusing on what's actually being said, rather than who's saying it.

"If language use evolved in service of face-to-face conversation ... There is no reason for language processing to develop an alarm mechanism that would continuously monitor the talker's identity and automatically signal a talker change," the researchers said. "Given the assumption of interlocutor stability, listeners are free to focus attention on the linguistic message."

"Change deafness" has its limits. In yet another experiment, the interviewer voice changed without warning from a woman's to a man's, and in this case eleven out of twelve participants noticed the change. "When talkers differ in vocal tract sufficiently, such as when talkers differ in gender, these bottom-up acoustic differences may grab attention even in the absence of top-down expectations," the researchers said.

ResearchBlogging.orgFenn, K., Shintel, H., Atkins, A., Skipper, J., Bond, V., & Nusbaum, H. (2011). When less is heard than meets the ear: Change deafness in a telephone conversation. The Quarterly Journal of Experimental Psychology, 64 (7), 1442-1456 DOI: 10.1080/17470218.2011.570353

Previously on the Digest: Phonagnosia - the inability to recognise people by their voice alone.

Post written by Christian Jarrett for the BPS Research Digest.


Andrew said...

So people miss a small, inconsequential difference, but notice a large one? Is that it?

Counselling Southampton said...

I know about change blindness but I never heard about change deafness. Now I have learned something new. Nice blog you got here
I agree, it's true that we humans don't really pay attention to small differences such as the changing of voice and that's because we focus too much about what the person on the other line was saying.

Web Designers said...

is that really it?i never known that there could be a sickness like that.. kinda weird to me.

Anonymous said...

Interesting article. But this line- "(the three women who played the role of interviewer across the different experiments had voice frequencies of 200Hz, 218Hz and 215Hz)." is confusing.

Cell phones can reproduce frequencies from about 400Hz to 7,000Hz. The human voice covers a wide range of frequencies and is constantly modulating. Female voices generally don't have much information in the 200-300Hz range.

Can you clarify? Please quote the original text because something has been lost in translation.

Unknown said...

Anonymous at 4.39 pm

Here are some relevant passages from the original journal article:

Two female experimenters conducted testing. The mean F0s for these experimenters, averaged over several randomly selected sentences, were 218 Hz and 200 Hz.

Three experimenters conducted testing. One of the experimenters, (F0 1⁄4 218 Hz) also conducted Experiment 1; the other two experimenters were new. The mean F0 for the two new experimenters was 202 Hz and 239 Hz, respectively.

<>I realise looking back at the article that I missed out one of the women involved - there were four in total so I've corrected that in the blog post.

>>A quick look on Wikipedia likely solves your query:

The voiced speech of a typical adult male will have a fundamental frequency from 85 to 180 Hz, and that of a typical adult female from 165 to 255 Hz.[1][2] Thus, the fundamental frequency of most speech falls below the bottom of the "voice frequency" band as defined above [for telephony]. However, enough of the harmonic series will be present for the missing fundamental to create the impression of hearing the fundamental tone.

Post a Comment

Note: only a member of this blog may post a comment.