Searching for the causes of scientific impact: Selection, social constructs, and schemata

In a previous post we have discussed the definition of scientific impact in what we called hub effect. Put simply, the hub effect in relation to scientific impact is defined by the number of people affected by the innovation a research group puts forward. The people affected by the innovation could be research peers who will use the innovation methods in their own research, or the public in general who will be either affected through the resulting technology (e.g., using an MRI, which results from basic research) or by changing concepts about their own daily lives (e.g., people who start exercising as a consequence of research demonstrating that this activity reduces the risk of cardiovascular disease).

But that was simply a statement of what scientific impact is, not what actually causes impact. Why should we care about what causes impact? Well, the assumption here is that researchers will want their innovation to affect as many people as possible, and having knowledge about the causes of scientific impact would enable them to more appropriately design research projects that have a higher likelihood to quickly spread and affect multiple people. From a BRIL perspective, knowing the factors associated with higher impact would allow us to better direct research resources toward projects that can lead to the greatest possible hub effect.

Although there is no one single answer to what causes impact -- impact is caused by different combinations of factors under different circumstances -- there seem to be a few frequent traits across innovation that is quickly spread.

First, innovation has to have an empirical basis to be sustainable in the long run. In other words, although marketing aspects immediately come to mind when talking about quick social spread, marketing alone about a research idea won't do it in the long run. This argument has been substantiated by multiple studies where selection was used as a criterion for how science evolves (Hull, 1990). For example, scientific peers will only spread the news about a certain innovation if it positively affects their own research. For example, if a new statistical method for survival analysis allows me to more accurately compare different treatments in terms of how long my patients will live, this innovation will certainly be spread across multiple studies. Also, if a new lab technique allows me to analyze blood samples to detect previously unknown markers of response to a cancer therapy, this method will certainly be tried across multiple diseases. In other words, the proposed innovation has to provide scientific peers with either the condition to see data that they could not see before, or to interpret data in a way that is different from what they had been able to up to that point in time. Using the terminology of biological selection, research innovation has to provide other researchers with a tool that will provide them with better fit to move their own research agendas forward.

Second, social factors such as power relations, networking, marketing effects, and funding, to mention a few, are also relevant regarding how a proposed innovation is received and spread. For example, research that is not well presented and marketed can go hidden for decades before it is discovered again, such as Gregor Mendel's discoveries about inheritance patterns. Also, being in a position of power or networking influence in a scientific society also facilitates the immediate spread of innovation as well as acquisition of funds that will enable the progress of somebody's research agenda. In sum, although social factors alone cannot sustain an innovation in the long run, science is a social endeavor that is conducted by researchers as a social group (Barnes, 1996).

Third, the spread of innovation relies on the perception that the innovation creates unexpected results. The key word here is expectation, in that researchers and the general public usually have a cluster of expectations regarding certain areas. Once those expectations are not fulfilled, a mix of emotions will rise, usually combining a mix of reactions such as interest, awe, despise and even disregard toward the innovation. Provided some of the previously mentioned selection and social factors are in place, an unsatisfied expectation increases the likelihood of creating a hub effect increases. Within cognitive psychology, this set of standard expectations is called schemata (Marshall, 1995). In a nutshell, schemata is the way by which we aggregate multiple associated concepts. For example, when a surgeon thinks of a hip fracture in an elderly person, the immediate association is with surgical treatment, prevention of post-operative complications, and long-term disability. Although the reasons why humans tend to aggregate concepts are unclear, it seems reasonable that schemata facilitate storage (the surgeon only has to think about hip fractures so that all other concepts immediately come to mind), synthesis (all factors are coherently linked, since fractures tend to lead to postoperative complications that then lead to disability), generalization (all elderly patients with a hip fracture will need measures to prevent complications and disability), and retrieval of similar experiences (whenever they see an elderly with a hip fracture, surgeons will automatically remember previous cases where complications and disability occurred). Schemata are relevant for research impact in that by changing a schema, a researcher changes the expectation their peers or the general public have about a concept. This unexpectedness can then lead to impact. As an example, once Mary Tinetti changed the schema about hip fractures to include the cause being a fall, a new plethora of factors were suddenly associated with the schema: Cognitive impairment, sensory problems in the lower extremity, slippery floors, walking toward the bathroom at night, among others.

Having at least three factors at hand -- selection, social factors, and schemata -- BRIL can now create a middle layer between research resources and the actual iterative process characteristic of research in the making. This middle layer will orient the project toward the largest possible impact. How these factors can be integrated into the underlying knowledge representation and reasoning behind BRIL will be the topic of a subsequent post.

Barnes, B., Bloor, D., & Henry, J. (1996). Scientific Knowledge: A Sociological Analysis. University Of Chicago Press.

Hull, D. L. (1990). Science as a Process: An Evolutionary Account of the Social and Conceptual Development of Science. University Of Chicago Press.

Marshall, S. P. (1995). Schemas in Problem Solving. Cambridge University Press.