Can you hear me now? Neuroscience paper sunk by audio stimulus error

Have you ever noticed that hearing something read aloud as you follow along helps you remember what you’re reading better?

Two bioengineers at Trinity College Dublin, Michael Crosse and Edmund Lalor, decided to investigate the underlying reason for the phenomenon. Unfortunately, after they published their findings in the Journal of Neurophysiology earlier this year, they tried to recreate the experiments and discovered that their equipment didn’t line up the audio and visual stimuli properly.

They did the right thing and contacted the journal for a retraction. Here’s the notice for “The cortical representation of the speech envelope is earlier for audiovisual speech than audio speech”:

We are retracting this article for the following reason: Following the publication of our manuscript, we aimed to replicate and extend our previous findings on audiovisual speech. Thus, we implemented a larger follow-up study with several different experimental conditions. In order to guarantee accurate synchronization of our audio and visual stimuli, we incorporated some additional electronic circuitry in our experimental setup. The result we reported in our original article did not replicate. This forced us to re-investigate the accuracy of the audiovisual synchronization used in our first study. On doing so, we detected a subtle yet consistent misalignment in the timing of our audiovisual stimuli. Thus, the latency shift we reported for audiovisual speech in the article cannot be trusted to be accurate. Latency shifts have previously been reported for discrete audiovisual speech in humans and for discrete non-human primate vocalizations. Whether similar latency shifts also occur in the context of continuous audiovisual human speech requires further investigation.

We offer our formal apologies for this error and for any inconvenience associated with the publication of the article. The paper is therefore being retracted by the American Physiological Society at our request.

The paper has yet to be cited, according to Thomson Scientific’s Web of Knowledge.

Hat tip: Richard Tomsett