Scientific innovation and the schemata cage

In an earlier post we mentioned that schemata can play a role in the impact associated with a certain research innovation. More specifically, we argued that schemata determine a set of expectations regarding a concept, and modifying or moving out of those schemata can lead to the perception of innovation. Although schemata facilitate storage, synthesis, generalization, and retrieval of similar experiences (Marshall, 1995), by creating clusters of concepts schemata also tend to lock researchers within a conceptual cage. In other words, because a group of concepts is tied together, it is difficult for the researcher or group of researchers sharing the same schema to foresee other possible schemata. This situation is troublesome when other alternative schemata could provide a better fit to explain or predict certain aspects of the behavior related to the phenomenon at hand.

But could we then ask researchers to step out of a schema so that the world can be perceived in a different way? There are several problems that would have to be overcome before a researcher can step out of a schema. First, schemata are frequently accessed without the researcher necessarily being aware of it. For example, in the classical example cited by Kuhn (1996), individuals seeing a deck with cards displaying black hearts and red spades most often didn't notice the inconsistency or were somewhat puzzled without knowing exactly what was wrong. It seems that because their previous schema for cards was to have red hearts and black spades, their perception was somehow impaired by this inconsistency with their pre-existing schema. The concept here is that pre-existing schemata tend to point attention to information that is consistent with the it while ignoring information that is inconsistent. This is referred to as a confirmation bias. In other situations inconsistent information is sub-categorized as an exception, leaving the original schema untouched. The latter is referred to as sub-typing, and is frequent when it comes to the evaluation of research theories. Two other factors tend to make us stay in the cage by increasing the access of previously existing schemata: salience and priming. Salience is the degree to which a particular element stands out relative to other elements in an environment. The higher the salience of an element, the higher the probability that schemata associated with that element will be accessed. For example, when a symptom is particularly characteristic of a disease, such as in the case of chest pain that is extended to the left arm for patients with heart attack (acute myocardial infarction), the mere presence of this pattern of pain is immediately associated with the possibility of a heart attack. Priming relates to any exposure immediately prior to a situation that caused a schema to be more accessed. The typical example is that, after the diagnosis of a rare disease, physicians tend to spend the next several months using that rare diagnosis as a differential for conditions that might have any of the symptom or signs associated with the schema for that rare condition. Another intriguing, but usually unnoticed example occurs very frequently in academic writing where, once the initial skeleton for manuscript is set, it is virtually impossible to restructure it escaping the initial framework. A recent episode, with detailed discussion among multiple participants, occurred in a group attempting to modify an entry in Wikipedia that the authors judged not to be adequate. Despite multiple attempts to move the structure away from its original content, the text still had the overall look of the initial structure even after several months (talk page on Charles Peirce).

Second, asking researchers to simply step out of a schema and look at the "raw data of the world" seems to be even more difficult. The facilitation associated with schemata in terms of storage, synthesis, generalization, and retrieval are not fortuitous: Schemata exist for the simple reason that our cognitive ability is limited, and schemata allow us to simply the representation of the world. It would be impossible for us to capture perceptions about the world in raw form. For us to be able to reason about the world, the raw data has to be somewhat packaged in concepts and categories so that we can represent the multiple facts before we can act. So, at least in our current state, stepping out of schemata is not an option, although future enhancements in knowledge representation might assist us in increasing the degree of detail of each category. I will deal with this issue in a future post.

The last option would then be to ask humans to simply change from one schema to another. This, I believe, is one of the possible ways to achieve innovation. The difficulty, however, is that this change cannot be made to a schema with randomly selected elements. Schemata are not a group of any elements hastily put together, but they have an internal rationale. In order to replace a schema for another, a rationale that is at least as strong as the precedent should be in place. This rationale goes back to the idea presented in the previous post post that, in a research environment where selection is present, the new schema should be able to either provide explanations or make predictions that were not possible through the previous schema. Selection is a process that requires multiple critical iterations, starting within the research group, then going through their immediate peers and, for concepts with a broad applicability, a criticism across different disciplines and the general public.

But, for now, let's focus on the initial, design phase of the research project, assuming that we have little control over the iteration that will occur later on, also assuming that our immediate concern is to establish conditions that are appropriate for changes in schemata. Are there any mechanisms that could help us get out of our schemata cages?

Psychology describes a few situations where schema boundaries can be overcome. For example, when a researchers' schema is in conflict with the social norms of a new group, they will be motivated to inhibit the influence of the original schema on their thinking and behavior. Whether researchers will successfully control the use of their old schemata is dependent on individual differences in their ability to exert self-control as well as the presence of situational incentives to maintain that control. All that said, when researchers stop consciously suppressing the influence of their old schema, a rebound effect can occur and the old schema might become hyper-accessible.

In real research situations, a situation of repression can occur in a variety of situations. Examples include inter-disciplinary activities involving the meeting of two or more groups with different methodological orientations, the inclusion of a single individual with a different training, or even when individuals go through cross-training. Other mechanisms are possible. For example, researchers could have brainstorming sessions before they dig into the literature that dominates a field, although this is obviously not a possibility when researchers are already experts in that field. Notice that here the idea is not that researchers will generate new schemata and then build their internal consistency from the ground up, but they will randomly pick other schemata that are already established and try to fit them into the concept. Because they are still not attached to the schema considered the official word on the topic, they have more freedom to consider other possibilities. In other words, they see other cages before deciding on which cage to jump into. Abbott, in a fascinating book describing the use of heuristics to generate innovation in social sciences, argues that these methods can be potentially effective in reaching innovation (Abbott, 1994). Finally, Pietrobon et cols (unpublished manuscript) has argued that the inclusion of polymaths into research teams is a possible method to include more ideas into the mix and therefore add to the potential of innovation.

Despite their initial appeal, at this point very little information is available on whether, when and where each of these approaches would be effective.


Abbott, A. (2004). Methods of Discovery: Heuristics for the Social Sciences. W. W. Norton & Company.

Kuhn, T. S. (1996). The Structure of Scientific Revolutions. University Of Chicago Press.

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