Learn how to keep away from widespread pitfalls when parsing circuit management of social conduct | Spectrum
Alan Lewis
Assistant Professor, Vanderbilt University
Autism researchers routinely rely on a number of methods to precisely control the activity of certain neural circuits in the brains of conscious animals. They have used these methods extensively to examine the basic neural control of social behavior and how changes in brain circuitry can contribute to the social difficulties characteristic of autism.
However, social behavior involves numerous measurable and non-measurable variables, making it difficult to identify causal relationships and achieve reproducible results across similar, but not identical, paradigms. For example, an ever-growing toolbox equips scientists with new methods of manipulating circuits that can open up new frontiers for experiments. However, a real mechanistic understanding of how these new methods differ from those previously used may have been several years ago, which in turn hampers reproducibility between laboratories and techniques. And editorial and financial priorities often reward claims that certain circuits have exceptional specificity for controlling higher-order social behavior, and lead scholars from dispassionate descriptions of how circuit manipulation affects behavior to narrative that might ignore evidence to the contrary .
These three complex factors are just examples of many of the challenges that fundamental and translational autism researchers should pause when conducting behavioral studies of circuit manipulation. What steps should be taken to avoid dubious conclusions and increase the likelihood of reproducible studies?
My colleagues and I addressed this concern in a recent article published in eNeuro. We have set forth simple, but often overlooked, practices and standards for designing and interpreting experiments for analyzing circuit control of behavior. This includes taking into account the many variables that could explain the results, using multiple behavioral tests, and reporting the results in a descriptive and unbiased manner.
Here I apply these practices briefly to the case of social behavior, as it is of paramount importance for autism research. None of these standards are revolutionary and all are suitable for use by any behavioral neuroscience laboratory. Because of their simplicity, these suggestions can facilitate generalization across laboratories and neuroscience areas to maximize the value of animal research.
Separate variables:
Imagine if you found that reducing the activity of a neural circuit decreases the time your experimental mouse spends interacting with another mouse in a behavioral task. It may be tempting to suggest that this circuit specifically “control” social interaction, but this conclusion would be premature.
This is because social behavior depends on numerous other intermediate behaviors or variables that can be influenced by manipulating the neural circuitry. For example, changes in locomotion, exploration, or sensory processing induced by manipulating the circuit could dampen social interaction without the circuit having any specific relevance to social behavior. More complex mediators are attention, salience, reward processing, fear, and motivation.
A creative researcher could identify an almost infinite number of intermediate variables that obscure the causal relationship between circuit activity and social behavior. That doesn’t mean researchers need to investigate each of these possibilities, or that a study that hasn’t been fully explored is too dubious to be of value.
In fact, the endless exploration of intermediate variables could further affect the reproducibility of scientific studies as researchers would have to subject their cohort of mice to many serial behavioral tests. Alternatively, testing many different cohorts of mice in a variety of assays could pose logistical, financial, and animal welfare challenges.
Therefore, it is in the field of autism and social neuroscience to establish a standard set of the main mediator variables that should be controlled in studies of social behavior. These standards need not be prescriptive if only one specific assay can serve such a purpose. Rather, an agreed set of controls would provide guidance to enable researchers to plan a course to accomplish what the field deems most important and avoid the need for randomly selected control experiments.
Describe behaviors:
The interpretation of experiments is, of course, influenced by what the researcher believes the experiment is measuring. For example, the observation that circuit manipulation shortened the time in which an experimental mouse examined another mouse in a neutral cage context could be interpreted as a social deficit compared to a control mouse. Another researcher with a different research goal, however, could interpret this result to reflect a faster familiarization with a new type of mouse. Both are reasonable and illustrate that a single performance measure cannot prove the functional nature of a circuit, requiring the use of multiple performance measures to support any assertion that a circuit serves a highly specific function.
Another easy way researchers can increase clarity and reproducibility between laboratories is to simply report what they have done and observed without interpreting those results. For example, the results portion of a study might say: “In a neutral cage, the interaction time between wild-type mice was longer than between knockout mice”, instead of “Social behavior was greater in wild-type mice than in knockout mice”. Researchers can then interpret the result in the discussion section, where it can also be contextualized with other behavioral experiments designed to examine intermediate variables in line with those considered important by the field.
The use of methods to manipulate the activity of neural circuits or to record their activity in vivo will continue to increase. These techniques offer tremendous opportunities to understand how the brain controls and codes social behavior, and how genetic or environmental variability associated with autism could affect social behavior. As neuroscientists, however, we need to be aware of the sometimes conflicting interests between the story that can be supported by data and the emphasis on specificity in circuit control of behavior rewarded by subsets of magazines or funding agencies. The simple steps discussed above will help move the field of autism and social neuroscience research toward reproducible circuit manipulation studies.
