Exploring the Citation Nexus of Life Sciences and Social Sciences

Which is more challenging: developing a vaccine for an unforeseen pandemic in under a year, or convincing people to take it?

A lesson from COVID-19 is that both the development and adoption of science can be equally challenging. From research to solution, societal factors lie in the last mile, capable of either facilitating or hindering the problem-solving power of science. Scholarly understanding of society, the realm of the social sciences, can complement life sciences by re-embedding biological entities into societal structures and navigating public acceptance of innovations.

To what extent then do social sciences influence life science research? We provide an estimate by tracing the citations made from ~170k life science papers in PubMed to social sciences. Using our sample, we estimate that 15-19 percent of life science papers cited social science knowledge, venues, or researchers at least once, resulting in 1.1-1.5 percent of references in 2018. Psychology accounts for roughly half of these citations; without it, the shares drop to 8-14 percent and 0.4-0.6 percent of references, respectively. Over time, we observe an increasing prevalence of citations to social science knowledge, with the percentage of life science papers citing social sciences rising from 12.1 percent to 14.7 percent from 2005 to 2018 and the proportion of references growing from 1 percent to 1.5 percent.

Fig.1: The growing prevalence of social science references in life sciences. Three classifications of science are adopted: cognitive classifications are derived by Microsoft Academic Graph (MAG) using publication texts, denoting the origin of knowledge; venue classifications are assigned to each journal by ScienceMetrix; organizational classifications denote the departmental affiliation of first authors, obtained from affiliation text in each paper. Panel (a) presents estimates accounting for all social science disciplines, and panel (b) excludes Psychology.

Percentages, however, only tell part of the story. At least as important is the question of how deeply social sciences influence life science research – do they merely touch the periphery or penetrate to its core? Using neural embedding models trained on how papers share references to infer cognitive location, we found that papers across diverse topics of life science research cite social sciences. For example, neuroscience research integrates economic concepts like prospect theory and risk perception with dopamine signaling in decision-making processes. Similarly, infectious disease studies use demographic indicators of household wealth to explore the socio-economic factors affecting infection patterns. Some social science methods, such as causal inference from econometrics, have taken root in life sciences and spawned life science-specific causal models. For instance, causal models for fMRI and attention networks are predominantly published in journals like the Journal of Neuroscience rather than journals specifically on econometrics.

Fig.2: The cognitive location of social science citers in seven journals (using venue classification and excluding psychology). Each dot represents a publication, with those citing at least one reference from social science journals highlighted as solid black circles. Colours denote clusters.

What kind of social science papers are cited in life science research? We found most cited social science publications rank among the top 10 percent impactful in their fields, however displaying varying degrees of interdisciplinarity. Some citations reflect a search for extramural knowledge, such as biological research citing information and library science on indexing or term weighting to study biomedical relationships. Conversely, others demonstrate transdisciplinary homogeneity, like the synergy between sociology and clinical medicine in understanding and evaluating patient care decisions, fostering clinical sociology.

Our analysis concludes by exploring why life science researchers cite social sciences, which we infer from the section where these references to social sciences appear. We believe these references predominantly serve contextual and non-empirical roles since they appear mainly in the introduction (41.5 percent) and discussion (39.7 percent). This is higher than the overall reference proportions for these sections, which are 37.5 percent for the Introduction and 36.9 percent for the discussion. Interestingly, social science references serve just as many methodological roles as average references, with 13.7 percent cited in the Methodology section, closely aligning with the overall reference proportion of 13.1 percent. Over 50 percent of references to social science methods and econometrics are found in the methodology sections, while that of information and library science, finance, and sociology amount to 24 percent to 34 percent.

The social and natural sciences are often perceived as two cultures, especially when it comes to research funding. However, our study provides empirical evidence of its spillover effect in science, though this role too has been recognized for decades. As early as 1958, Harvard University initiated a fellowship program to train medical professionals in social sciences. In recent decades, considerations of ethical, legal, and social issues have become crucial for research in human genetics and infectious diseases.

The growing synergy between life sciences and social sciences is achieved not only by life sciences becoming more social but also by social sciences’ increasing interdisciplinarity. We expect this trend to continue with rising public engagement and attention to research, coupled with real-time policy needs and the implementation of science.