Pigeons definitely get a bad rap. Some might consider them mere rats with wings, purveyors of pestilence, distributors of dung, but rock doves aren’t, well, as dumb as their name might suggest. Pigeons are perhaps the world’s most accurate homers, they seem to have an innate knack for game theory and they can detect breast cancer in mammograms better than many doctors.
So when researchers in Germany reported in 2017 that pigeons were as adept, if not better, than people in multitasking, the findings seemed plausible. The study, which appeared in Current Biology, garnered a bit of media attention, including this piece in The Scientist, and has been cited five times, according to Clarivate Analytics Web of Science.
Turns out, that was a flight of fancy.
According to the retraction notice:
In our Correspondence, we reported evidence leading us to conclude that pigeons are on par with humans when tested with a behavioral task that demands simultaneous processing resources; in particular, we claimed that pigeons show faster responses than humans when sub-tasks are separated with a short STOP–CHANGE delay of 300 ms—the “SCD 300” condition (time advantage of 200 ms). We have subsequently discovered, however, that the MATLAB script that was used for the analysis of reaction times in the pigeon paradigm was wrongly indexed. Therefore, the measured SCD 0 and SCD 300 reaction times from the paradigm were randomly assigned to the corresponding conditions in the analysis script. The subsequent data analysis was thus based on erroneous reaction times of both conditions. The error was detected when a software update made a change in the analysis script necessary. When the correct script is used for the analysis, we find that pigeons and humans show comparable reaction times in both the SCD 0 and SCD 300 condition. The reported time advantage of pigeons compared to humans no longer reaches statistical significance. Thus, in Figure 1C the black bars indicating the reaction times of pigeons are erroneous, while the reaction times of humans (white) remain correct. The same applies to the Supplemental Information: in Figure S2B the black bars indicating the reaction times of pigeons are erroneous, while the reaction times of both human experiments (white and striped) remain correct. We are therefore retracting the paper and apologize to the scientific community for any inconvenience.
We emailed the lead author for comment but didn’t hear back.
Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on Twitter, like us on Facebook, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].
Scripts just like excel sheets for data analysis are a mess.
More often wrong than right.
http://theconversation.com/the-reinhart-rogoff-error-or-how-not-to-excel-at-economics-13646
We need to find something more reliable.
There are fields where computerized data analysis is unavoidable. But yes, it should always be accompanied by a look at the raw data. And a test run or two of the code, please! Basing conclusions on a code one person, often a junior scientist, has written and which has not been checked by anybody else is just not good enough.