1 Introduction

Is scientific knowledge mainly an extension and specification of everyday knowledge, or does science require or lead to a break with common sense intuitions? If there is a discontinuity, how is it possible to learn science at all? To what extent do common-sense intuitions enable or limit scientific reasoning? The answers to such questions have important implications for philosophy of science and science education. The aim of this paper is to explore the potential for cross-fertilization between the discussions about conceptual change in these two domains, motivated by the claim made by Hoyningen-Huene that systematicity theory offers a particularly suited platform in order to investigate the relation of the sciences and common sense.

Hoyningen-Huene’s thesis is that scientific knowledge can be characterized as an extension of everyday knowledge, distinguished by an increase in systematicity:

Science develops out of common sense of the respective historical time or out of a nonscientific knowledge practice due to an increase in systematicity. Thus, we can determine the relationship between science and common sense by investigating what the effects of this increase in systematicity are, first upon common sense itself and later during the ensuing scientific development (Hoyningen-Huene 2013, p. 187)

By clarifying how science grows out of common sense, historically and in contemporary scientific practice, Hoyningen-Huene emphasizes continuity between everyday knowledge and scientific knowledge. Yet, he stresses an important difference in the degree of systematicity in nine dimensions in science, compared to non-scientific activities. The nine dimensions are: (1) Descriptions, (2) Explanations, (3) Predictions, (4) The defense of knowledge claims, (5) Critical discourse, (6) Epistemic connectedness, (7) An ideal of completeness, (8) Knowledge generation, and (9) The representation of knowledge. The main thesis is that science—compared to everyday knowledge about the same subject—more carefully considers and excludes possible alternative explanations, samples more systematically, makes a more extensive recording and evaluation of data, and that scientific knowledge has a higher degree of connectedness to other knowledge claims.

Hoyningen-Huene contends that the difference between everyday knowledge and scientific knowledge is a difference in degree, but he departs from the so-called continuity-thesis in emphasizing that scientific knowledge often comes to deviate from common sense and our everyday experiences as science develops. He stresses that many scientific insights are inaccessible to direct perception or even conflict with what we experience. For instance, whereas the Aristotelian tradition of physics can be seen as a successful specification of common sense, he argues that the same cannot be said for modern physical theories involving claims about anti-matter, string-theory, relativity, universes in multiple dimensions, space being bent etc. Hoyningen-Huene considers the implications of such theories as “slaps in the face of common sense” (Hoyningen-Huene 2013, p. 193). A further claim is that giving up common sense notions and intuitions is a price that we sometimes have to pay for the increased systematicity of science. That is, common sense is sometimes a “victim of science”, resulting from the incumbent increase in systematicity (Hoyningen-Huene 2013, p. 188, 194). These claims raise important questions about the extent to which “common sense” can be said to be a precondition for or a victim of science, and about the role of systematicity in the move from common sense to scientific knowledge.

Other papers in this special issue critically examine the strength of what Hoyningen-Huene calls systematicity theory, particularly the attempt to demarcate scientific knowledge from everyday knowledge with reference to systematicity. My aim is to highlight important cognitive aspects that are not given much attention in Hoyningen-Huene’s account. Hoyningen-Huene primarily deals with continuities and breaks in a comparison of the endpoints of conceptual development, i.e. similarities and difference between scientific and everyday knowledge.Footnote 1 With this choice he leaves unspecified many aspects associated with the cognitive reasoning processes involved in learning and doing science. I stress how the “bird’s eye” perspective of systematicity theory can gain from a finer-grained analysis of conceptual development in science education and philosophy of science.

It may be objected to the framework presented that drawing parallels between conceptual change in science and science education is of limited use due to differences in the levels of organization. For instance, Grieffenhagen and Sherman (2008) have argued that drawing analogies between Kuhn’s view of conceptual change in science (at the level of a community of scientists) and science education (at the level of individual students) is misleading. The analogy certainly has limitations, particularly the way that Kuhn’s incommensurability thesis has been taken up to describe learning processes in science education (Sect. 2; see also Levine 2000). Importantly, however, the comparison of literatures on conceptual change in the two domains does not have to rely on a Kuhnian view of conceptual change, or to assume that conceptual change only operates at specific levels in the two contexts (see also Sect. 4). The comparison made in this paper is not dependent on a Kuhnian view of science, but is simply an attempt to explore fruitful connections in the two literatures that can help nuance common assumptions about the relation between common sense and science. I shall, however, return to Kuhn in Sect. 4 to comment on the connection that Hoyningen-Huene himself draws between systematicity theory and Kuhn’s account.

We begin with an introduction to debates in science education and cognitive science about what common sense and scientific reasoning entail (Sect. 2). This will serve as a background for reexamining Hoyningen-Huene’s assumptions about common sense and scientific knowledge, and for comparing similar debates about conceptual change and heuristic strategies in philosophy of science (Sect. 3). Both literatures stress important roles of common sense intuitions in scientific reasoning that may only become visible through a finer-grained examination of the knowledge process. While such studies provide resistance to assumptions made about discrepancies between science and common sense, Sect. 4 examines situations where there seems to be a greater distance between common sense intuitions and scientific theories. In such contexts, I explore a possible role of systematicity theory in specifying the nature of external support required for learning and doing science. Specifically, I explore the potential of Hoyningen-Huene’s account to clarify institutional aspects of science through a more contextualized notion of systematicity. Section 5 offers concluding remarks.

2 Perspectives on common sense from science education

One of the important aims of philosophy of science is to reflect critically on the hidden assumptions of scientific practice. But philosophy of science should be equally critical of its own assumptions when discussing philosophical questions such as the similarities and differences between science and common sense. Assumptions about what common sense is, and about what learning and doing science entails, greatly influence the perceived relations between common sense and science. Moreover, the grain of analysis (temporal perspectives as well as scope of analysis) can influence the conclusions drawn about the relation between these. To illustrate these issues, I start by introducing three theories about learning and common sense in science education (the Theory Theory, the Ontological View, and Knowledge in Pieces, respectively).Footnote 2 All three accounts draw parallels between scientific reasoning of novices and experts. But whereas the two former emphasize that learning science entails abandoning common sense “misconceptions”, the latter stresses that expert knowledge is only possible on the basis of intuitive knowledge. The comparison to the Knowledge in Pieces account is also interesting because this account explicitly describes the difference between intuitive and expert knowledge as a difference in the degree of systematicity (diSessa 1993b, Systematicity section). I highlight the practical implications of each theory while relating these to the approach taken by Hoyningen-Huene.

2.1 Learning science: unlearning common sense

The Theory Theory is a prominent theory for conceptual change in science education that draws on developmental psychology and cognitive science. The version of the account that I shall focus on here argues that children learn through a process similar to the way that scientists reason through hypothesis generation and theory revision in response to evaluation of evidence (Carey 2009; Chi 1992; Gopnik et al. 1997, 1999). The description of theory revision is inspired by work in philosophy of science, but proponents of the Theory Theory argue that the analogy can be read both ways. That is, studying cognitive development can help us understand how science works as well as vice versa. Accordingly, this view is sometimes called ”the child as scientist view”, “the scientist in the crib” or “the scientist as child” (Solomon 1996).Footnote 3

The Theory Theory emphasizes continuity in the reasoning processes of early development and scientific research in the sense that “there are common cognitive structures and processes, common representations and rules, that underlie both everyday knowledge and scientific knowledge” (Gopnik and Glymour 2002, pp. 117–118). Gopnik and Glymour provide a version of the Theory Theory that draws on the representational framework of causal maps, particularly Bayes nets (directed graphs), to specify how theory change in both domains implies a recovery of an “accurate causal map of the world” (Gopnik and Glymour 2002; Gopnik and Wellman 2012). A causal map is an abstract representation of the causal, correlational and counterfactual relations between objects and events in the world. Gopnik and Glymour argue that also everyday or ‘folk’ theories have the character of causal maps and posit coherent relations among different events and objects.

Hoyningen-Huene’s description of scientific knowledge as distinguishable from everyday knowledge by a higher degree of systematicity seems to be in alignment with Gopnik and Glymour’s description of causal maps in the two domains, but Hoyningen-Huene’s account is not restricted to the narrow notion of knowledge as causal maps. There is much more to science than representing the causal structure of the world, and Hoyningen-Huene elaborates on some of these aspects when clarifying the dimensions of prediction, knowledge generation, the representation of knowledge, and the role of generalizations in physics and chemistry. Hoyningen-Huene’s and Gopnik’s accounts thus focus on very different aspects of science and have complementary virtues, but also complementary limitations. While Hoyningen-Huene’s account ignores many psychological and social aspects of scientific practice, Gopnik’s account may overemphasize the sufficiency of developmental psychology for providing an epistemic epistemology (see, e.g., Solomon 1996). This point is important for evaluating claims made by the Theory Theory considering the revision of the ‘theories’ when faced with conflicting evidence.

For the Theory Theory, the continuity of cognitive structures and conceptual development does not extend to the relation between the “naïve” theories and scientific theory. According to the Theory Theory, learning science involves a process of theory revision and transformation according to the strength of evidence. Gopnik and Glymour (2002, p. 129) for instance argue that whereas adults in everyday life are rarely forced to revise their causal knowledge, the situation is very different in cognitive development and science education. In their view, both child development and science learning involves abandoning prior (intuitive) theories that otherwise would block conceptual change because these provide a false picture of the causal structure of the world. Moreover, the naïve ideas (or theories) are taken to be systematically intertwined and coherent, akin to scientific theories, which means that giving up intuitive notions and conceptions is often a hard challenge. Accordingly, some proponents of the Theory Theory therefore talk about learning as a series of personal scientific revolutions, drawing on Kuhn’s notion of “local incommensurability” (Carey 1985, 2009; Kuhn 1962/1996).

The Ontological View puts even more emphasis on the challenge to go beyond naïve theories, arguing that prior ontologies impede progress in students’ learning (Chi and Slotta 1993; Chi 2013; Reif and Larkin 1991). Chi for instance emphasizes how scientific ontologies are categorically distinct from common sense ontologies. According to this account, the implication for science education is that teaching science must involve the development of new and directly instructed ontologies. That is, the naïve ontologies or theories are considered “false” and need to be replaced by “correct” ones. In this view, learning is equal to “changing naïve conflicting knowledge to correct knowledge” (Chi 2013). Thus, learning involves abandoning or bracketing prior naïve knowledge because the naïve ideas conflict with the development of new ontologies that better support scientific theories. Accordingly, Wiser aims to develop instructional designs that protect students from the prior common sense notions, providing a “free-standing” network of ideas that does not borrow from or interfere with naïve theories (Wiser 1995, p. 34). Rather than smoothly extending out from the student naïve ideas, here perceived as “misconceptions”, new theoretical frameworks must be built almost from scratch (Slotta and Chi 2006, p. 286).

These views seem to conflict with Hoyningen-Huene’s emphasis on continuities and differences primarily in the degree in systematicity between everyday knowledge and scientific knowledge. As we shall see in the following, the Knowledge in Pieces (KiP) theory allows for more flexible pre-scientific elements of reasoning and can support this aspect of Hoyningen-Huene’s account. At the same time, the KiP theory suggests a constructivist approach to learning for which learning is only possible through connections to already established knowledge. The KiP theory thus raises the question whether common sense can meaningfully be considered a “victim of science”.

2.2 Learning science: constructing from current knowledge

Whereas the Theory Theory and the Ontological View claim that students cannot make progress without abandoning prior ontologies and intuitions, constructivist theories of learning hold that knowledge can only develop by building on current knowledge. In this context, the notion of ‘constructivism’ implies a rejection of the traditional “transmission view” of teaching and learning in which objective knowledge is passively transmitted to students and added to their existing knowledge structure. From a constructivist perspective, new knowledge is always constructed at the basis of previous experiences (diSessa 1993a, b; Knorr-Cetina 1981; Piaget 1974; Posner et al. 1982). The claim that accepting scientific knowledge requires giving up common sense intuitions must therefore confront questions about how learning and doing science should be possible at all, if not drawing on common sense intuitions.

What motives a constructivist approach to learning? As noted in the previous section, much research in science education has focused on strategies to replace the ‘misconceptions’ of students with new scientific explanations or ontologies. However, empirical studies of learning situations show that the so-called ‘misconceptions’ are very robust. For instance, the beliefs held by participants in special conceptual change programs compared to control groups display no significant differences (Lijnse 2000). Similarly, longitudinal studies involving interviews with a group of students over several years demonstrate the continuous influence of ‘life-world knowledge’ shaped by early experiences on reasoning about ecological processes (Helldén and Solomon 2004). This suggests that the different knowledge domains are not incommensurable and that common sense intuitions are not given up (see also Sect. 4).

Detailed knowledge on how current belief systems of students (or scientists) interact with and accommodate new ideas and findings is still lacking. But steps towards the development of a finer-grained analysis are currently taken in both science education and philosophy of science. The KiP theory illustrates how a different view may result from a more contextualized ‘microgenetic’ analysis of student learning (diSessa 1993a, 2014a, b, forthcoming). The notion of microgenetic analysis in this context refers to the fine grain of detail of data collected during a real-time process of conceptual change (see also Siegler and Crowley 1991). Consider as an illustrative example a recent study by diSessa where a real-time learning situation was observed (diSessa 2014b, forthcoming). In the study, students in a high school class construct a model of thermal equilibration at the basis of experiments with hot and cold water in a test tube. The aim of the exercise is for students to get an understanding of Newton’s description of temperature change as proportional to the difference of temperatures among two objects, but the students were not given any prior theoretical information. The study shows how students were able to draw on intuitive or “primitive” ideas about the behavior of the liquids as anthropomorphic agents and still arrive at the intended learning goal (see diSessa 2014b, p. 813). For the Theory Theory and the Ontological View, the attribution of agency to liquids or molecules would be considered a “misunderstanding” that needs to be replaced with “correct” theories and ontologies to enable scientific reasoning. In contrast, diSessa argues that such “misconceptions” constitute part of scientific learning and understanding (see also Gupta et al. 2014). In the example, the description of the temperature changes in terms of preferences of liquids as agents facilitate conceptualization about what happens to the liquids outside equilibrium, and the students spontaneously reformulate their findings to arrive at a formal description of how the rate of temperature change is driven by temperature differences.Footnote 4

The possibility embraced by the KiP theory for a partial reliance on common sense intuitions in scientific reasoning should be understood against the background of an important difference in the underlying assumption about our belief systems of the different theories. Whereas the Theory Theory and the Ontological View emphasize that naïve knowledge make up a coherent system of beliefs, KiP posits a system of many loosely connected knowledge elements called “phenomenological primitives”, or p-prims (diSessa 1993a, b). p-prims provide generic causal schemas for making phenomena and relations sensible to us, such as descriptions of causal processes in anthropomorphic terms. Another example is what diSessa (2014a) calls “Ohm’s p-prim”: the idea of a direct relation between agentive “effort” and some kind of result, e.g. between effort in running and speed. Changing the view of naïve theories from being misconceptions to p-prims opens for the possibility to see these as essential in scaffolding learning processes (diSessa 1993b, 2014a, b).

Rather than viewing common sense intuitions as something that get in the way of scientific reasoning, the task becomes to understand in which contexts they work and in which they are of limited help. diSessa (1993b) illustrates how different p-prims are “activated” in different situations as a result of “high cuing-priority” of some p-prims (and not others), depending on the context. For instance, diSessa (2014a) points out that Ohm’s p-prim provides an intuitive understanding of situations where greater effects give rise to greater results, but face limitations when the task for instance is to understand Newton’s law of inertia stating that motion perpetuates itself if not acted upon by a force. In such cases, however, Ohm’s p-prim is not “unlearned” but put aside in favor of other ideas that may better scaffold the learning. For instance, inertia may be understood against the background of opposing situations in which students draw on “constraint p-prims” (diSessa 1993b, p. 121) that allow students to understand unconstrained movement in comparison to constraints experienced by movement in water versus air.

Compared to the Theory Theory and the Ontological view, the developmental histories of the cognitive elements are considered more independent, affording a greater flexibility of naïve knowledge in learning situations. Rather than abandoning initial conceptual frameworks, this involves a process in which the student gets a feeling (intuitively or more explicitly) of the contexts in which specific frameworks are useful. Similarly, Mortimer (1995) has argued that learning science does not necessarily involve conceptual change but changes in the student’s conceptual profile. From this perspective, learning science in part becomes a way of exploring the fruitfulness of conceptual tools that are already at hand for understanding new observations but also in knowing when the tools meet their limitations and need to be supplemented or “displaced” (diSessa 2014a; see also Helldén and Solomon 2004). Finer-graining the analysis thus opens a space for the use of “misconceptions” as access-points to scientific theories, rather than only obstacles to progress. Moreover, the emphasis of the flexibility of p-prims and the development of more systematic knowledge systems raises an important question about whether scientific knowledge is developed from cognitive systems that do not display the same degree of systematicity and coherence as the products of science (diSessa 1993b).

In summary, whereas the KiP theory seems to support Hoyningen-Huene’s emphasis on a continuity between common sense and science, it is difficult from a constructivist perspective to make sense of how common sense could be a “victim” of more advanced research (Hoyningen-Huene 2013, p. 188, 194). Against this challenge it may be objected there are important differences between the learning situations in science education and the cognitive demands for navigating in the conceptual spaces of theoretical physics that we do not have intuitive perceptual access to. I shall return to Hoyningen-Huene’s examples of “breaks” with common sense in Sect. 4. First, I highlight connections to the points taken up by diSessa and recent discussions on conceptual change and heuristic strategies in philosophy of science.

3 Philosophy of science and biased heuristics

As the previous section illustrates, the idea that theories are revised through rejection of particular conceptual frameworks or ontologies has been contested by researchers in science education. Recently, philosophers of science have taken issue with the ways in the Theory Theory has been justified with reference to work in philosophy of science. For instance, Glennan (2005) characterizes the view of the Theory Theory on conceptual development through theory revision as an odd mixture of Popper and Kuhn. On one hand, conceptual development is taken to happen through a process akin to Popper’s idea of falsification of hypotheses in view of conflicting evidence. But this process is described through a radical replacement of common sense ontologies akin to a revolution in a Kuhnian sense. This combination is problematic because, unlike the Theory Theory and the Ontological view, Kuhn (1962) objected to the view of the history of science that pictures earlier theories as naïve, mistaken or underdeveloped precursors to more correct ones. Moreover, Kuhn questioned Popper’s simplified ideal of falsification with reference to commitments to a set of values, theories and practices in normal science. In contrast, the Theory Theory and the Ontological view give the impression that evidence uniquely determines theory choice. This view leaves only a marginal role for the influence of theoretical frameworks required for any observation, as well as of social and institutional factors (Solomon 1996; see also Sect. 4).

In summary, the Theory Theory does not seem to find strong support in recent empirical work in science education, or in Kuhn’s account. Moreover, modern philosophy of science has moved beyond the traditional view of theory testing in examining how scientists reason with models as mediators or epistemic artifacts (Morgan and Morrison 1999; Knuuttila 2011; Nersessian 2002, 2008). Glennan describes the difference between the Theory Theory and the model building account in the following way: “On Gopnik’s account, theories face the tribunal of experience directly, while on the modeler’s account, theories only face that tribunal as mediated through models” (Glennan 2005, p. 224). The important difference is whether scientific products (as causal nets, models or theories) stand in direct representational relations to the structure of the world or whether this relation is mediated via models. In Glennan’s view, theories are too abstract to be directly consistent or inconsistent with experience or data. When models are tested, they are not falsified in the simplistic sense because models are not true or false. Glennan therefore proposes that the Theory Theory would stand better if drawing on recent work in philosophy of science on model building. If, however, the Theory Theory buys into this assumption, theory revision can no longer be understood as simple mismatches between a real-word target and a mental causal model (see also Nersessian 1996; Solomon 1996). This also has important implications for the view of the relation between common sense and scientific knowledge in the context of scientific research.

The aforementioned aspects call for greater attention to how scientists (and science students) evaluate evidence and model explanations, but also of how new knowledge is generated. Whereas previous philosophical accounts took hypothesis-generation to be a random creative endeavor outside the scope of philosophy of science (e.g., Popper 1959), philosophers have recently attended to the important roles of common heuristic strategies that reduce the unfamiliar scientific operations to more familiar tasks. Heuristic strategies span from the use of mundane analogies and metaphors to more specialized strategies such as particular diagrammatic representations or the mechanistic strategies of decomposition and localization common in molecular biology (Bechtel and Richardson 1993/2010; Hesse 1963; Keller 2002; Nersessian 1995). Similarly, exploration has recently received attention as yet another strategy of scientific inquiry (see Steinle 1997 in relation to experiments, Gelfert 2016 in relation to models, and Shech 2015 in relation to idealization). Similar to the emphasis on common sense as leverage for scientific reasoning in science education (Posner et al. 1982), philosophers and cognitive scientists have argued that common sense intuitions and analogies play important roles also in advanced scientific reasoning (Carruthers 2002; Nersessian 1992).

These aspects are, however, easily overlooked from a “bird’s eye” perspective examining similarities and differences of knowledge products. From this perspective, the relation between common sense and science may appear as discontinuous, for instance that “common sense knows very little or nothing about bacteria living in symbiosis with us” (Hoyningen-Huene 2013, p. 194). Considering the specialized theories of microbiology and (the lack of) counterparts in everyday knowledge, it is indeed difficult to see parallels in the two domains. However, although the size and amount of microbial symbionts may be alien to common sense, it may be the case that the understanding of symbiotic relations in biology can usefully be facilitated through our notions of collaboration in social contexts. Similarly, diagrams of molecular factories or lock and key models are often used to leverage the understanding of biological mechanisms, although these misrepresent the biological reality. It is often assumed that analogies and metaphors mainly are useful as pedagogic tools in science education but play a minor role in scientific reasoning of practicing scientists. Such views are, however, challenged by empirical work in philosophy of science demonstrating that cognitive strategies such as analogies, metaphors, thought experiments and mental modeling are essential to scientific reasoning (Dunbar 2002; Nersessian 1992, 2008; see also Green 2013). Drawing on many years of observational studies of practicing scientists, Dunbar even argues that mundane analogies are among “the most frequent workhorses” of the scientific mind (Dunbar 2002).

Similar questions thus arise in the context of philosophy of science. Do common sense reasoning processes, such as analogies from the social or engineering domains to living systems, enable or impede scientific reasoning? And to what extent is it possible to do science without resources from common sense?

3.1 Reasoning with “false” models

A debated issue in philosophy of science is the extent to which the function of heuristics—to simplify the problem space of scientific analysis—enables or limits conceptual development and progress in science. The points of criticism can be illustrated with the debate in philosophy of biology on the use of engineering approaches and optimality assumptions.Footnote 5 Analogies between designed systems and organisms have a strong impact on development of mental models, and biologists often benefit from the ready-made design language to conceptualize biological functions. If more is known about the design space for specific functional capacities in engineering than about the mechanism underlying a functional capacity in biology, analogical reasoning can enable a transfer of modeling tools or specific hypotheses from engineering to biology. Nevertheless, some have argued that engineering analogies and metaphors impede progress in biology because they oversimplify the problem space for analysis and make researchers blind to other relevant alternatives. Specifically, engineering approaches have been criticized for drawing a misleading parallel between intentional design and natural selection as “optimization” (for an overview of the debates, see Green 2014, 2015).

Similarly, Pigliucci and Boudry (2011) argue that machine and design metaphors are “bad for science and science education”, and should be avoided in order to minimize misunderstandings. Examples referred to involve the idea of the genome as a program, the brain as a computer, and the view of organism as a machine consisting of static parts (see also Boudry and Pigliucci 2013; Nicholson 2013). Likewise, the usefulness of reverse engineering approaches in biological research has been questioned with references to the plasticity and dynamic nature of the structure and function of living systems compared to machines (Braun and Marom 2015). The call to avoid certain heuristics in science education and scientific research resonates with the view of the proponents of the Ontological View when arguing that:

Teachers should not try to “bridge the gap” between students’ misconceptions and the target instructional material, as there is no tenable pathway between distinct ontological conceptions... Indeed, students’ learning may actually be hindered if they are required to relate scientifically normative instruction to their existing conceptualizations (Slotta and Chi 2006, p. 286).

Importantly, however, the question about the productive and problematic aspects of research heuristics reaches far beyond the issue of the truth-value of the assumptions they rely on. As diSessa (2014a, b) argues in the context of science education, it would be a category mistake to discuss whether common sense notions are true or false. Rather, they are reasoning tools that work in some circumstances and not in others. Similarly, the use of analogies and metaphors in scientific reasoning should not be evaluated as empirical statements about the world, but on the extent to which they productively guide these. “False models” are fruitful means for knowledge generation if systematically calibrated with independent sources of evidence (Wimsatt 2007), and design heuristics in biology can be productive despite the reliance on false assumptions if the analysis can uncover to what extent the systems are similar and different (Knuuttila and Loettgers 2013). Still, it is important to further investigate whether these enforce certain metaphysical assumptions that frame our ontological and scientific viewpoints. That is, the debate about biased heuristics raises the deeper question about the extent to which we can become prisoners of our own abstractions so that these impede us from seeing and establishing new relevant connections (Levins and Lewontin 1985).

Thus, whereas some have argued that the biases of research heuristics can be accounted for through testing of the generated hypotheses, this strategy seems insufficient if the problem lies in the missed opportunities for conceiving other—and better— alternatives (or in the lag time for reaching alternative options). For instance, relying on comparisons between intentional design and optimization of traits via natural selection may lead to a neglect of the influence of other evolutionary forces than natural selection (Green 2014). The Ontological View therefore seems to have its merits in reminding us that we view the world through a certain theoretical lens that may affect the way that we evaluate and make sense of evidence. But the Ontological View takes it too far in the claim that learning requires students or scientists to completely abandon common sense ontologies and start reasoning with new ones. This may be abandon tools that are useful or even necessary for scientific reasoning.

Just like abandoning common sense ontologies in the context of science education may not be a feasible option, it is relevant to examine the extent to which scientific reasoning is possible at all without relying on common sense. Even scholars arguing against the use of engineering metaphors and functional or ‘teleological language’ in biology have problems articulating their research results without these (Krohs 2015). Similarly, even when accepting theories that conflict with our perception of the world (e.g. special relativity), we do not seem to give up common sense intuitions (Mortimer 1995). When common sense intuitions turn out to be counterproductive to scientific reasoning, they may be displaced (or “nudged out”) rather than replaced (diSessa 2014a). That is, some intuitions may be bracketed when reasoning within certain problems spaces while remaining functional in other contexts. From this perspective, one can rephrase the claim of machine metaphors as being “bad for science and science education” as a point about the constraints on their scope when applied in biology. Following Wimsatt (2007), we may say that successful reasoning with biased heuristics involves an awareness of the limitations of our epistemic tools and their context-dependent uses, rather than a replacement of false models with correct ones.

In Sect. 2, we saw that diSessa (2014b) questions the assumption of the Theory Theory and the Ontological view that intuitive knowledge make up a systematic and coherent, and therefore inflexible, network of knowledge elements. Similarly, observational studies of scientific practice suggest that “causal reasoning in science is not a unitary, cognitive process, as argued by Gopnik and Glymour [...], but a combination of very specific cognitive process[es] that are co-ordinated to achieve a causal explanation” (Dunbar 2002, p. 157). Dunbar emphasizes how causal reasoning in science involves dozens of different problem-solving techniques, suggesting that scientific thinking contains many different interrelated sub-components.Footnote 6 Thus, whereas increasing systematicity may be a hallmark of scientific reasoning and scientific knowledge, the lack of systematicity (in the sense of coherence) of the intuitive knowledge may be what facilitates the use of cognitive elements in many different contexts. It may also support the survival of common sense notions even when seemingly counterintuitive theories, such as special relativity, are accepted. To discuss this question further, we now return to the hard cases where the distance between common sense and science appears to be great.

4 The systematic organization of science

I begin with some general reflections on the emphasis on continuities or discontinuities of common sense and science in the current debates. As pointed out by Gopnik and Glymour (2002), some of the different viewpoints may simply result from differences in perspective—reflecting a choice of focus on either the similarities or the differences, or on a specific grain of analysis. That is, more or less dramatic discontinuities between science and common sense will be visible, depending on the ways in which the topic is framed and investigated. Importantly, however, such choices have practical implications.

On one hand, emphasizing the difference between common sense and science may underestimate the ways in which scientific reasoning relies on everyday cognitive processes and overestimate the difference between the cognitive processes of expert and novices (diSessa forthcoming). Attention to the methodological framework and the underlying assumptions is therefore of key importance. For instance, expert-novice studies have highlighted the fluency of expert reasoning and limited reliance on common sense notions compared to novices. Yet, critics have argued that the comparison is based on problem-solving cases that for the experts do not go beyond routine puzzle-solving. If so, the cases fail to give a realistic picture of how scientists deal with ill-defined problems in research and are of limited use as guidelines for science education or (cf. diSessa, forthcoming; Gupta et al. 2010, 2014; Schauble and Glaser 1990).Footnote 7 Rather than accepting the categorical assumptions about naïve and expert causality—e.g., that agent causality is inappropriate or unnecessary for reasoning in physics—a more contextually grounded analysis is needed to identify the situations in which such strategies facilitate or impede scientific reasoning. That is, we must examine scientific reasoning using methods that account for situated cognition (Nersessian 1992, 1995). As argued, finer-grained analyses suggest a more central and less problematic role of common sense in science education and scientific research.

On the other hand, it may be objected that empirical studies of how common sense notions guide specific reasoning practices in science education cannot be uncritically transferred to contexts where scientific theories appear to be particularly counterintuitive. Hoyningen-Huene (2013) distinguishes between three different shades of deviation from everyday knowledge, namely (i) knowledge that results from a specification of common sense knowledge, (ii) new knowledge that is unrelated to common sense knowledge, and (iii) new knowledge that breaks with common sense (Hoyningen-Huene 2013, pp. 190–192). As examples of the latter he mentions explanations in theoretical physics, such as special relativity or string theory. Considering these examples, it may be argued that emphasizing a straightforward continuity between common sense and science may underestimate the difficulties for many to learn theories that are distant from everyday experiences. Particularly, it may underestimate the extent to which science is dependent upon external support, formal education and cultural factors (Carruthers 2002; McCauley 2011). In other words, attention to the differences may help clarify why some aspects of science are particularly challenging to accept and what is needed to support these processes.

4.1 Science at the frontiers of common sense

It is commonly known in cognitive psychology and science education that ideas that differ radically from established knowledge can be very difficult to learn (Ausubel 1963; Helldén and Solomon 2004; Posner et al. 1982). Modern science often traffics in highly abstract representations that require a lot of work to accept and master (McCauley 2011). While attention to evidence is an indispensable prerequisite for early cognitive development as well as for doing science, the assessment of data and evidence in the two contexts may be distinguished by the extent to which interpreting data requires particular training. To understand evidence in advanced science disciplines, we often need a “theoretical lens” that cannot easily be acquired. Stanovich (1999) stresses that navigating in science and in the modern technological society requires decontextualized reasoning skills, and that: “for intellectuals to argue that the ‘person in the street’ has no need of such skills of abstraction is like a rich person telling someone in poverty that money is really not important” (Stanovich 1999, p. 200).

Hoyningen-Huene (2013) points to the history of science in highlighting the difficulties of accepting new theories with greater distance to common sense. As examples he mentions the timeframe of acceptance of the Copernican picture in physics and Darwin’s theory of evolution through natural selection in biology. Likewise, McCauley (2011, p. 108) observes that although microorganisms were discovered already in 1674, it took nearly two centuries to take seriously the idea that these tiny living systems could affect something as disproportional as human health. The way that insights from science continue to shock and puzzle us indicates that many aspects of modern science require a break with naïve realism. At the same time, however, there appears to be historical changes in what is considered common sense. As Hoyningen-Huene contends, although we are unable to perceive the speed and rotation of the Earth around the sun, we today accept the Copernican picture as a fact (Hoyningen-Huene 2013, p. 192). Thus, connections between common intuitions and science are not fixed once and for all, and common sense resources may be characterized by some degree of adaptive flexibility.

When Hoyningen-Huene argues for a break with common sense, he points to situations where science tells us that the natural world is not as it appears to us. Aside from the Copernican theory that “deprived the celestial motions of their objectivist status”, Hoyningen-Huene (2013, p. 195) mentions how phenomena such as smells and colors became subjective “secondary” qualities, and how we today have to accept wave-particle dualism. Einstein’s special theory of relativity is in Hoyningen-Huene’s view an example of how common sense can be a victim of an increase in overall systematicity, because the common sense notion of (absolute) simultaneity is rejected. Special relativity is thus “a kind of knowledge that directly contradicts common sense knowledge” (Hoyningen-Huene 2013, p. 191).

Not surprisingly, researchers in science education has pointed to the same example as one of the particularly difficult learning challenges for college students and also highlight the conflict between ordinary and scientific metaphysical beliefs (Posner et al. 1982). Interestingly, however, Posner and colleagues observe how students in their empirical study draw on analogies to established knowledge and experiences to make special relativity more accessible as piecemeal accommodations of the new theory. To describe the final step of accommodation, Posner and colleagues draw an analogy to Kuhn’s account of conceptual change in science, which has been criticized (Grieffenhagen and Sherman 2008; Levine 2000). However, it is interesting to note that they describe accommodation as a gradual and piecemeal affair where established metaphysical beliefs are often protected from rejection. Similarly, Knobe and Samuels (2013) report from a set of experiments based on questionnaires that both scientists and non-scientists draw on a conception of innateness considering biological traits that is influenced by moral judgments. Depending on the framing of the questions, both groups are however capable of ‘filtering out’ their initial intuitions and to use a more scientific approach.

The aforementioned empirical studies raise an important question about whether the common sense notions are ever completely given up, or whether they are just bracketed in some situations (as discussed in Sect. 2.2). As Grieffenhagen and Sherman (2008) observe, scientists do not seem to replace everyday language with scientific language to describe such phenomena in daily life. Although we accept and understand that the perceived movement of the sun is an apparent motion and that space has no absolute direction, we still speak of the sun rising in East. This suggests that common sense notions survive as compartmentalized notions that continue their lives in other contexts. Thus, we should perhaps take a more pluralistic stance and view the different conceptual schemes as tools for understanding different perspectives of reality, where some dominant common sense notions must be bracketed to allow for reasoning within new possibility spaces.

One example of how subjects may bracket certain aspects of common sense resources is illustrated by a study in which the responses of students shift in response to different prompts that draw on what the researchers categorize as i) life-world concepts or ii) as physical terms (Helldén and Solomon 2004). The students appear to shift between different “knowledge domains” and languages rather than adopting a mixture of life-world and physical terms. In response to similar studies and results, some researchers in cognitive science have suggested that reasoning is conducted by two different systems (Evans and Over 1996; Evans and Stanovich 2013; Stanovich 1999, see also Kahneman 2011; McCauley 2011). I shall not go into the details of such accounts or the empirical support for these. However, I wish to highlight that their emphasis on a slower and rule-based system for decontextualized or reflective cognition needed for scientific reasoning is suggested as an explanation for the difficulties of learning science. Because reflective cognition requires greater effort, learning and doing science requires a great level of external support and needs to be systematically trained (e.g. McCauley 2011). Whereas Hoyningen-Huene’s notion of systematicity does not have much to offer on the finer-grained cognitive aspects of scientific reasoning, it is worth exploring whether the account can provide insight to methodological and organizational systems that provides support for scientific activities.

4.2 Possible implications for science education

In this section, I explore the relevance of Hoyningen-Huene’s and also Kuhn’s account for science education. I begin by considering objections to using these accounts on the characteristics of science as a source of insights for science education. I have already mentioned objections to using Kuhn’s incommensurability thesis in the context of science education. Grieffenhagen and Sherman (2008) argue that “Kuhn’s conceptual schemes are tied to what scientists do (e.g., performing new experiments and calculations). However, school pupils are not engaged in work in that sense, since they do not produce new explanations of the natural world. Pupils are in school to learn what others have discovered” (Grieffenhagen and Sherman 2008, p. 22). In their view, learning science is better seen as a refinement of everyday knowledge (as, perhaps ironically, Einstein himself stated). Moreover, they stress that whereas science education focuses on the cognitive development of individual students, Kuhn’s account was focused on how science functions on the level of a community (see also Sect. 1).

I have acknowledged the limitations of viewing conceptual development in science education in analogy with scientific revolutions (Sects. 2 and 3), and activities in the two domains differ. Yet, as clarified in Sect. 3, there appears to be many overlaps in the cognitive strategies used by scientists in exploratory phases and strategies for “active learning” in science education (Gelfert 2016; Nersessian 1995; Shech 2015; Steinle 1997). Science education should not only be about learning theories. As numerous empirical studies show, deep learning is better facilitated by “inductive teaching” where students actively work on solving problems (Prince 2004). Moreover, pointing to science as a refinement of common sense does not clarify why learning some aspects of science often is particularly challenging and what can be done about this. Hoyningen-Huene’s account offers a balance between the different views in the literature by on one hand acknowledging how science grows from common sense by an increase in systematicity, while also pointing to examples where the connection to common sense seems more complex and partially discontinuous. Greiffenhagen and Sherman may argue that these discontinuities are only seen in the generation of new and highly abstract knowledge in science. Yet, as we have seen, the difficulty of reasoning in the framework of special relativity is a challenge not only in science but also in higher education (Posner et al. 1982).

Moreover, science education can also be approached from a community perspective that goes beyond the psychology of individual student. The relevance of Kuhn’s account on these aspects should not be dismissed.Footnote 8 What Hoyningen-Huene views as the most fruitful aspects of Kuhn’s account, also for the comparison to systematicity theory, is not the description of paradigm shifts but the features associated with normal science.Footnote 9 Specifically, Kuhn (1959, 1962) clarifies how the training of students to follow paradigmatic solutions (exemplars) to well-described research problems enables students to solve new problems. Hoyningen-Huene (2013, pp. 163–165) explicitly acknowledges the connection to this aspect of Kuhn’s account with respect to generation of new knowledge (dimension 8). He argues that exemplars provide systematic guidance for normal science in the sense that they provide a specific orientation towards paradigmatic problems. Hoyningen-Huene thus seems to suggest that systematicity can clarify and extend existing accounts on how students are trained. I contend, however, that the notion of systematicity would be more useful for science (and arguably also for philosophy of science) if made more concrete and more contextualized. In the following, I motivate this claim and provide some suggestions for new paths to explore.

4.2.1 Potential gains from a more contextualized account of systematicity

Hoyningen-Huene’s (2013) main aim is to clarify what in general distinguishes science from everyday knowledge, and the call for a more contextualized account may seem to miss the point of his project. However, it is debatable whether the abstract nature of systematicity succeeds in providing even “some tenuous sort of unity” (Hoyningen-Huene 2013, p. 169). Hoyningen-Huene explicitly acknowledges that systematicity can be realized in countless ways. This context-sensitivity stems from the variety and historical context of scientific disciplines and practices. Hoyningen-Huene gives a few examples of how systematicity may play out differently in different disciplines due to different aims (e.g., concrete vs. generalized descriptions and explanations), and he admits that no less than nine dimensions of systematicity are needed to respond to counterexamples questioning the difference between everyday knowledge and science (Ibid, pp. 209–210). The pursued level of generality and abstraction of the notion of systematicity raises important concerns about whether the term has much substantial content (Ibid, p. 179), e.g. whether existence of possible counterexamples would be likely (Ibid, p. 169). When writing that “scientificity is a notion that is extremely dependent on the various discipline and time” (Ibid, p. 206), one may wonder whether systematicity has merely become a synonym for science, particularly when it is admitted that systematicity theory offers “nothing concrete that all the sciences will have in common” (Ibid, p. 169). Accordingly, several scholars have argued that the flexibility of the notion of systematicity that Hoyningen-Huene sees as a virtue is, in fact, a vice because the theory is too vague and flexible to be rejected even in principle (Rowbottom 2013).

While I have given Hoyningen-Huene’s book a more charitable reading than Rowbotton’s (2013) review, I agree that “HH tends to be at his best when he’s working at the level of the trees, rather than that of the woods”. I see an unexplored potential in a more contextualized analysis of how the regulative ideal of increasing systematicity is played out and prioritized differently in different fields. To illustrate how systematicity is realized, Hoyningen-Huene often provides rather detailed examples, for instance when describing the characteristics of systematic representations in different fields such as graphical representations in mathematics, the periodic system and formulas in chemistry, and evolutionary trees and mechanistic diagrams in biology (Hoyningen-Huene 2013, Sect. 3.9). Through such descriptions, systematicity theory reminds us that learning science is not only about “learning what others have discovered” (Grieffenhagen and Sherman 2008, p. 22) but also about committing to certain epistemic standards, representational methodologies and systems of knowledge.

Systematicity theory seems well suited for specifying the characteristics of different systems of knowledge or different practices, by pointing to the specific dimensions for which specific disciplines prioritize their systematic endeavors (collection of big data, systematic sampling and classification, criteria for statistics or coherence etc.). It may be argued that my suggestion would make systematicity instrumental to norms in science, rather than an end in itself (cf. Hoyningen-Huene 2013, Sect. 5.3). While I acknowledge Hoyningen-Huene’s intention to provide a general description of the difference between scientific and everyday knowledge, I believe that systematicity theory could potentially reach a broader uptake if explored as a tool for the comparison of standards in specific situations and in different traditions. For instance, rather than a regulative ideal implying that more advanced science is characterized by a higher degree of systematicity, different concrete instantiations of systematized knowledge systems and practices can help clarify important aspects of what Kuhn (1959) described is the “essential tension” between commitment to the well-established methods and theories and the call for new knowledge in science. Similarly, Andersen (2013) points to the “second essential tension” between disciplinary tradition and interdisciplinary innovation. Since diverging epistemic standards often impede interdisciplinary collaboration (e.g., Rowbottom 2009; Green et al. 2015), attention to what is systematized, how and why may be more fruitful way to draw on systematicity theory.

For instance, systematicity theory could help us understand not only how exemplars provide “systematic guidance” in science but also how these are established and learned. A more contextualized notion could potentially be connected to discussions of systematicity in science education that also points to a connection to Kuhn’s notion of examplars (diSessa 1993b). diSessa has argued that: “collecting and systematically attaching p-prims as distributed encodings for physical principles” can account for “a structural and knowledge-based view of the process that Kuhn identified as central to learning a discipline, the process by which students learn to see the exemplar outside its initial context while problem solving” (diSessa 1993b, p. 145). He argues that intuitive knowledge elements facilitate the constitution of exemplars, such as the harmonic oscillator, by a gradual clustering and organization of p-prims into what he calls distributed encodings. p-prims, distributed encodings, exemplars and advanced scientific theories are thus levels in the reasoning process that may be characterized as increasingly systematic but also increasingly context-dependent due to the rigidity of the increasingly coherent knowledge system. The KiP theory is thus one suggestion for how the increase in systematicity may be given a more concrete and dynamic aspect by studying how specific intuitive notions can interpolate between familiar phenomena and the highly schematized abstractions of advanced science. Moreover, the KiP theory complements some aspects of Hoyningen-Huene’s account in examining different dimensions of systematicity such as mutual use and plausibility of different p-prims (diSessa 1993b).Footnote 10 At the same time, Hoyningen-Huene’s account unpacks some dimensions of systematicity in science that are not accounted for in cognitive analyses in science education.

Systematicity theory is a useful reminder that doing and learning science is not only about establishing and learning scientific results, respectively. It is also about establishing and understanding the systematized structure and the epistemic and social norms that stabilize the scientific enterprise, whether understood as a disciplinary matrix (Kuhn 1962) or an epistemic culture (Knorr-Cetina 1999). As Hoyningen-Huene points out, the organization of scientific practice and knowledge is often overlooked by philosophers of science. Attention to these issues is, I would argue, also important for science education because learning science is also about navigating in more systematically constrained methodological and theoretical spaces. Examples are the ways in which students are trained to more carefully consider and exclude alternative explanations when presented with problems to solve (Hoyningen-Huene 2013, p. 71), to more systematical sample, record and evaluate data (Ibid, p. 83), and to reflect upon and minimize sources of error (Ibid, p. 89). Similarly, at university level students may be taught to give different weight to different sources of evidence, as exemplified in the evidence hierarchy in medicine (Ibid, p. 102, 197).Footnote 11 These institutional frames may greatly influence the ability of students to adopt a scientific and systematic approach. To explain the persistence of value-laden conceptions of innateness despite scientific training, Knobe and Samuels emphasize that what distinguishes scientists from ordinary folks is not a different ontology but that the “behavior of scientists is molded by characteristic features of their external situations” (Knobe and Samuels 2013, p. 84). Among these they mention how scientists are typically confronted “with situations that encourage them to think systematically about a whole range of cases” (Ibid).

Organizational aspects of science support a more rigorous criticism and defense of knowledge claims (dimension 3) that is characteristic of higher education as well as science. Although operating on a different level (Grieffenhagen and Sherman 2008), science education is ideally not only about teaching students scientific knowledge but to learn to use abstract representations, make mathematical derivations, reflect upon arguments, establish connections to different pieces of knowledge and compare evidence. These are what Stanowich calls decontextualized reasoning skills and what McCauley (2011) argues are in need of external support as they need to be trained. The systematic methodologies exemplified by Hoyningen-Huene may help understand how these skills become institutionalized in both science and science education. Moreover, a more contextualized approach to systematicity may help us understand the implications of increased systematicity in different contexts. For instance, increasing systematicity in experimental biology may result in increasingly detailed descriptions and explanations, whereas increasing systematicity in theoretical biology inspired by engineering and physics may consist in abstraction from such details for the purpose of generalization (Rowbottom 2009; Green et al. 2015). In other words, a more contextualized notion of systematicity can not only help us go beyond the focus on theories but also help us understand how systematicity plays out in different contexts.

Attention to the finer-grained cognitive aspects of conceptual change and iterations between everyday and scientific reasoning could also help substantiate some of the claims made about the role of systematicity for theory choice and broader claims about the role of systematicity in scientific development (Hoyningen-Huene 2013, Sect. 5.3). Considering Einstein’s special theory of relativity, Hoyningen-Huene (2013, p. 193) argues that the change of perspective resulted from “the realization that a relativized notion of simultaneity would increase the coherence and thus the systematicity of explanations for some class of phenomena”. Thus, Hoyningen-Huene considers a more systematic account necessary to “obtain maximal coherence of all the relevant data”. Hoyningen-Huene does not account for the details of the process of conceptual change, and against this background it is difficult to examine whether increased systematicity really explains the most salient features of the shift. The historical case seems much more complex. Examining the role of systematicity in such contexts requires closer attention to the process of conceptual development. The comparison of the end-products of different historical periods in science tends to neglect the continuity and finer grained aspects of the process of conceptual development. In the context of Hoyningen-Huene’s specific example, the development of relativity theory should also be understood in the historical context of work on electromagnetism and atom theory that makes the discontinuity to Newtonian mechanics less apparent (Nersessian 1992). Thus, while the notion of systematicity has potential for clarifying important difference between everyday knowledge and scientific knowledge (and between different scientific practices), bridging the gap to cognitive analysis in science education and HPS can help substantiate and nuance many of the claims made in Systematicity: The Nature of Science.

5 Concluding remarks

Hoyningen-Huene (2013) argues that systematicity theory is particularly well suited for clarifying the relation between common sense and scientific knowledge. Yet, by focusing on rather abstract and static conceptual categories (knowledge products and general categories of activities in everyday and scientific practice), he leaves unspecified many aspects associated with the reasoning processes involved in learning and doing science. Hoyningen-Huene points to gaps between ordinary and scientific knowledge on selected topics, e.g. that hermeneutics certainly is “alien to common sense” (2013, p. 193) or that common sense has little to say about microbiology. Yet, because the comparison is made on this general and rather abstract level, the important role of common sense notions to make these aspects intelligible is largely neglected. That is, the degree of discontinuity is sensitive to the grain of analysis. The general comparison says little about whether giving up common sense intuitions is required for the piecemeal process by which we accept or learn scientific knowledge. What both constructivist theories of learning and the cited literature in philosophy of science highlights is that we must not only analyze the state of differences between common sense and scientific knowledge, but also the change itself from pre-scientific to scientific ideas and methods. Accordingly, I have argued that while the “bird’s eye” perspective in Systematicity can gain from a more contextualized cognitive analysis.

I have argued that fine-graining the analysis reveals a more prominent role of common sense in scientific reasoning. Whereas Hoyningen-Huene argues that giving up common sense is a price we sometimes have to pay for increased systematicity, empirical studies in science education suggest that common sense are very robust and survive the scientific theories that supposedly break with these. To clarify this robustness, despite the acceptance of scientific theories, scholars have suggested that our cognitive systems and common sense notions (or p-prims) are more flexible and compartmentalized than previously assumed. This raises interesting questions about whether systematic scientific knowledge is made possible by a lower degree of epistemic connectedness (dimension 6) of common sense. Accordingly, we may question whether an increase in systematicity is always a good thing or even an end in itself (cf. Hoyningen-Huene 2013, Sect. 5.3). Systematicity in the nine dimensions affords more efficient and robust analyses, but it does so by relying on some degree of methodological and theoretical rigor. This rigor may provide tensions between tradition and innovation (Kuhn 1959), and between disciplinary standards and interdisciplinary innovation (Andersen 2013). Attending to such issues, I have suggested that there is an unexplored potential of systematicity theory in a more contextualized analysis of how students become familiar with the organizational aspects of science, and how increasing systematicity is realized and prioritized in different fields.

It may be objected that my suggestion is at odds with Hoyningen-Huene’s aim to determine what in general distinguishes science from everyday knowledge. However, systematicity may not be most useful as a notion describing the general difference between science and everyday knowledge, but in specifying how scientific reasoning and collaboration are externally supported by systematically structured activities and institutions.