Behavioral research at Journal of Neuroscience

In a recent issue of Journal of Neuroscience, the editors described an ongoing policy of increasing the percentage of papers that are “desk-rejected”, that is, rejected without peer review. The data showed that from Jan. 2014 to March 2016, the percentage of such decisions increased from 5% of submitted papers to around 25%. What concerned me was that the policy appeared to focus on a specific group of neuroscientists: those who primarily relied on behavioral experiments to infer function of the brain. We conducted a survey to quantify this impression.

130 people responded to the survey, 62% of whom had published a paper in JNeurosci. These authors represented about 342 articles in JNeurosci (this is a lower bound because a handful of authors had published 10 or more articles but the survey counted their contributions as 10). Of the papers that the authors had published in JNeurosci, about 40% were behavioral, without explicit neural correlates. Therefore, the survey participants represented a group of scientists that relied substantially, but not exclusively, on behavioral research.

The participants reported that they had experienced 148 desk-rejections. Among the desk-rejected manuscripts, about 75% were rejected because the editors felt that the work “lacked insights into neural mechanisms”. Here are some examples of the letters sent to the authors:

• “For purely behavioral studies, our criteria have evolved to require that a behavioral study provides novel insights into the underlying neural representations and mechanisms.”

• "The study does make some experimental predictions, but they are all behavioral."

• "while the study had a number of strengths, the emphasis was on behavioral processes rather than neural mechanisms which are the focus for The Journal of Neuroscience.”

• “It is, unfortunately, the case that it is not possible for us to consider manuscripts where the emphasis is on behavior.”

•  “I am afraid that it has become much rarer for behavior only manuscripts to be sent out for review at The Journal of Neuroscience.”

This policy has left a strong impression on the participants, as noted in their comments.

• "It is well-known that it is useless to submit behavioral studies to this journal even if neural mechanisms are discussed in the Introduction and Discussion."

• "To understand how the brain works, we have to understand the structure of the information processing and predict the behaviour before jumping to see the neural correlates of it. Therefore, I believe behavioural research is a fundamental part of neuroscience. It is sad to see one of the journals I respect is underestimating the value of behavioural neuroscience."

• "Although my personal experience included neural data, I definitely share the impression.  Quite a few journals including j neurosci are rejecting papers at the editorial stage and I have a sense behavioral studies are being excluded more; also agree that this is not wise for the journal."

• "I gave up submitting papers to JN or reviewing for them, a position which I also advertised to colleagues who came to me asking for advice whether they should consider JN as an outlet for their own work. My feeling has become that submitting papers that are mostly behavioural/computational to JN is a waste of time and effort."

• "The loss of behavioral research is dangerous to neuroscience. The idea that we will understand the brain without a nuanced understanding of behavior, i.e. the only neural output that ultimately matters, is misguided."

• "It will be a huge mistake to give any kind of low priority to behavioral/psychphysics/cognitive only studies! These papers are the basics of neuroscience. After all, this is what we aim to understand : human (or animal) behavior! Moreover, even as an electrophysiologist, i would say without hesitation that the contribution of purely behavioral papers to our understanding of computations in the brain is huge, and equal to that of invasive studies. Sometimes it is even more convincing to draw a conclusion about the neural correlates by observing the behavior, than by observing the actual neural correlate. this is due to two main factors: 1. Limitations and technical confounds of the physiological measures; 2. The superiority of a clean targeted well-thought-of purely behavioral design, compared to when compromised when one need to accommodate technological confounds of physiological methods. To sum, behavioral and psychophysics is, almost by definition, the foundation of neurosciences."

Fortunately, the editors are beginning to respond to these concerns. Senior Editor Dan Sanes left a constructive comment noting that this issue will be discussed at the journal's board meeting in November. Senior Editor Yavin Shaham spoke with me and noted that a policy statement that appeared to focus on behavioral research was problematic. Editor-in-Chief Marina Picciotto mentioned that she has asked for data regarding desk-rejections in the past 6 months and will compare them to similar data from 3 years ago to ascertain whether behavioral work has been unfairly impacted.

Meanwhile, David Herzfeld has begun a systematic, quantitative analysis of all papers published in the history of JNeurosci. The data has allowed us to quantify the citation impact of behavioral research, with the results published here.

In a thoughtful perspective, John Krakauer, Asif Ghazanfar, Alex Gomez-Marin, Malcolm MacIver, and David Poeppel considered the question of whether behavioral experiments are fundamental to advancing neuroscience. They noted that whereas the focus of neuroscience appears to have shifted to neural circuits, behavioral experiments provide the guiding vision of what that circuit might be doing. They wrote: "when scientists ask 'how does the brain generate behavior,' they are in fact asking a question best approached through behavioral work, specifically task analysis, aided by theory, that allows behavior to be decomposed into separable modules and processing operations... The neural basis of behavior cannot be properly characterized without first allowing for independent, detailed study of the behavior itself."