Disruptive evolution: harnessing functional excess, experimentation, and science as tool Teppo Felin 1,2, * and Stuart Kauffman 3 1 Department of Strategy and Entrepreneurship, Huntsman School of Business, Utah State University, 3500 Old Main Hill Logan, Utah 84322, USA. e-mail: teppo.felin@usu.edu, 2 Department of Strategy and Innovation, Saïd Business School, University of Oxford Park End Street, Oxford OX1 1HP, United Kingdom and 3 Institute for Systems Biology, 401 Terry Ave N, Seattle, Washington 98109, USA. e-mail: stukauffman@gmail.com We appreciate the thorough feedback and comments provided by Luigi Marengo and Rouslan Koumakhov. This article has also benefited from interactions with colleagues involved in the University of Oxford’s “Foundations of Value and Values”-initiative and workshops—including conversations with George Ellis, Cecilia Heyes, Colin Meyer, Denis Noble, David Sloan Wilson, and many others. We also appreciate useful feedback and email communication regarding this paper from Peter Richerson. This paper also benefited from comments from the organizers, panelists, and audience members at the 2022 Academy of Management symposium on “Evolutionary approaches to innovation: New developments.” Thanks to Amy Felin and Jack Leonard for carefully proofreading multiple versions of this manuscript. The usual disclaimers apply. *Main author for correspondence We explore the limitations of the adaptationist view of evolution and propose an alternative. While gradual adaptation can explain some biological and economic diversity, it cannot account for radical innovation (espe- cially during the past 10,000 years). We argue that ubiquitously available but dormant “functional excess” provides the raw material for evolutionary disruptions. Harnessing this excess requires directed experimen- tation and what we call “protoscientific” problem solving. We highlight the implications of these arguments for evolutionary theory, including evolutionary economics and strategy. JEL classification: C63, D8, D83, L1, L2, P00, B25 1. © The Author(s) 2023. Published by Oxford University Press in association with Oxford University Press and the Industrial and Corporate Change Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com Introduction Evolution has been modeled as a process of gradual hill-climbing on a fitness landscape. Any trait, feature, or capability is essentially seen as the product and outcome of gradual adaptation and long-run selection (e.g., Dawkins, 1997; Cosmides and Tooby, 2013; Lloyd, 2015; Dennett, 2017). This adaptationist logic has been applied to various biological phenomena, including genes, organisms, features, and traits, as well as cognitive processes (e.g., Wright, 1967; Maynard- Smith, 1978; Godfrey-Smith, 2001; Orzack and Sober, 2001). The adaptationist logic has also been central in evolutionary economics. Specifically, the central construct of this literature— routines—is the habits or “habitual response[s]” of managers and organizations to “problems of the past” (Winter, 2004: 164; cf. Hodgson and Knudsen, 2010). In adaptationist fashion, routines are said to best be understood “by reference to the evolutionary process that has molded them” (Nelson and Winter, 1982: 16). In the field of strategic management, hill-climbing and environmental adaptation are also central mechanisms for explaining the evolution of strategy and technology (e.g., Kauffman, 1993; Levinthal, 1997; Kauffman et al. , 2000). Downloaded from https://academic.oup.com/icc/advance-article/doi/10.1093/icc/dtad057/7332166 by University of Oxford user on 31 October 2023 2 T. Felin and S. Kauffman While gradual adaptation is important, it cannot explain the explosion of biological diversity and innovation—particularly over the past 10 000 years 1 —that are evident from the evolution- ary record through to the modern economy (e.g., De Beaune, 2004; Benton and Emerson, 2007; Read and Van Der Leeuw, 2008; Arthur, 2009; Kauffman, 2016; Bar-On et al. , 2018; Gatti et al. , 2020; Grinin et al. , 2020). Explaining this explosive growth and innovation—including varied niches, tools, technologies, institutions, and ways of “making a living”—is especially puzzling in the context of human evolution. Innovation seems pervasive, even though human cognition is said to be “maladapted,” “lagging,” and “mismatched” to the environments that humans find themselves in (e.g., Cosmides and Tooby, 1994b; Richardson, 2010; Lloyd et al. , 2011; Stanovich, 2011; Brown and Richerson, 2014; Li et al. , 2018; Workman and Reader, 2021). We are told that “our modern skulls house a stone age mind” (Cosmides and Tooby, 1997: 85; also see Pinker, 1997: 42). 2 Yet, despite the supposedly stone age mind of humans, we are surrounded by deliberately constructed human-built innovations, technologies, and institu- tions that can scarcely be explained by appealing to a gradual or random process. These radical leaps and disruptions seem to contradict Darwinism’s fundamental tenet of evolution, “natura non facit saltus”—that is, “nature does not make leaps” (Dennett, 1995: 288; also see Hodgson, 2004). The central question then is: how do we explain and account for leaps and “disruptions” in evolution? 3 We argue that functional excess and the human capacity for a type of scientific reasoning provide a plausible explanation. To foreshadow our specific argument, we argue that problem solving and experimentation enable biological organisms—including human actors—to harness dormant functional excess. To set the stage for these arguments, in the paper, we first point out how existing evolutionary explanations of function are largely backward-looking, algorithmic, and reductionist (Dennett, 1995). Specifically, extant evolutionary arguments are focused on reverse engineering and isolating a single, independent, proper function—the so-called “normal- proper,” historical function (Neander, 2017)—for any given trait, feature, or object (see Griffiths, 1993; Millikan, 1993; Dawkins, 1997; Pinker, 1997; Klein et al. , 2002; Dennett, 2017; Eaton, 2020). Using the metaphor of a toaster as an example (from Cosmides and Tooby, 1994a), we show how this form of evolutionary theorizing succumbs to what we call “functional essen- tialism.” We point instead to the ubiquity of (dormant) functional excess across biological and economic domains. Functional excess provides the raw material for evolutionary disruptions. However, the harnessing of functional excess is not automatic. We argue that directed experi- mentation and what we call “protoscientific” problem solving provide plausible mechanisms for explaining innovation in the context of evolution. We build on a very specific strand of biol- ogy to make these points (e.g., Uexküll, 1926; Riedl, 1984; also see Felin et al. , 2014). Beyond evolutionary theory, we also discuss how our arguments extend, link to, and have critical impli- cations for evolutionary economics and strategy—including the central concepts of habit and routine. We point toward a form of evolutionary pragmatism and highlight how this suggests a broader conception evolution, one that can more readily account for evolutionary leaps and changes. 1 Ten thousand years is an extremely short interval of time in evolution and thus we cannot point to genetic evolution to account for these radical innovations and changes. 2 Steven Pinker similarly argues that the “modern mind is adapted to the Stone Age.” Specifically, he argues: “Our brains are adapted to that long-vanished way of life, not to brand-new agricultural and industrial civilizations. They are not wired to cope with anonymous crowds, schooling, written language, government, police, courts, armies, modern medicine, formal social institutions, high technology, and other newcomers to the human experience (1997: 42). 3 The idea that humans are cognitively maladapted to present environments does not hold water in many domains, particularly where most of the modern environments (including technologies and institutions) we encounter are human- made and designed. Instead of arguing that human cognition lags the demands of present environments, one might argue the opposite: human cognition is far too good at addressing present problems (perhaps at the expense of future ones). For example, humans can readily, say, over-fish (or even completely deplete) lakes and oceans using various technologies. In a sense, humans can quickly exceed their carrying capacity and become “super-predators”—completely extinguishing resources like their prey (cf. Efferson, 2008). To ensure this does not happen, humans (sometimes) purposefully delimit or ban the use of certain technologies, or put in mechanisms and institutions to ensure that valuable resources are not depleted and that different (or more sustainable) resources are secured. Downloaded from https://academic.oup.com/icc/advance-article/doi/10.1093/icc/dtad057/7332166 by University of Oxford user on 31 October 2023 Disruptive evolution 3 2. Evolution and reverse engineering function: a review and example Perhaps the most general way of conceiving of evolution is as a gradual process of adaptation, where the environment selects things—traits, features, or objects—for their functional fitness. Any object we might focus on (say, a heart or a limb) is there for a reason, because it was “good for something” (Dennett, 2017: 52), because it enhanced fitness and survival. The focus on function is at the very heart of evolutionary theory. As Cosmides and Tooby argue, “natural selection theory is a theory of function” (1994a: 96; also see Dawkins, 1997; Dennett, 2017; Pinker, 2021). The argument is that the traits, features, and objects fit past environments by enhancing survival, enabling them to “make a living” (Dawkins, 1986)—in short, they solved particular problems and thus were selected for the functions they performed well. Unfit traits and features are selected out. As summarized by Neander, the “core idea of these theories is that the function of an item is to do what it was selected [by natural selection] to do” (2017: 1148). Any improvements are necessarily gradual, as environments select for traits that enable better fitness over long-term, evolutionary time frames. The central method for making these adaptationist claims about function has been the use of reverse engineering (see Williams, 1966; Cosmides and Tooby, 1994a; Dennett, 1995, 2017; Pinker, 1997; Neander, 2017). As argued by Dennett, the “reverse engineering perspective is ubiquitous in biology” (Dennett, 2017: 54; Richardson, 2010). Reverse engineering means that to explain a given trait or feature we need to retrospectively and historically look at its primary function: what it was “designed” by evolution to do, why it was selected, and what particular evolutionary problem it solved. The critical assumption here is that focal traits are “isolable” in terms of their function—because “without this readily discernible isolation, reverse engineering in biology would no doubt be humanly impossible” (Dennett, 1995: 216; also see Duchaine et al. , 2001). The idea of reverse engineering function is critical for understanding evolutionary arguments, though the concept of function can be somewhat abstract. Thus, to illustrate the idea of reverse engineering function in evolution, Cosmides and Tooby (1994a) introduce the metaphor of a “toaster” (for other examples and metaphors of evolutionary functional analysis, see Buss, 1995; Dawkins, 1997; Dennett, 2017; Williams, 1966; for example, see Klein et al. , 2002 for a closely related discussion of a “three-hole punch”). They argue that the first goal of this type of func- tional analysis is to understand what a given item (in this case a toaster) is, what it was “designed to do—what its function is” (Cosmides and Tooby, 1994a: 96; for similar examples, see Den- nett, 2017). Cosmides and Tooby argue that to properly engage in rigorous functional analysis, we need to be laser-focused on the toaster’s primary function, rather than focusing on what it “ can do” (Cosmides and Tooby, 1994a: 95). This distinction—between the primary versus pos- sible function of an object—is critical to evolutionary analysis, and also highlights our point of departure. To illustrate the importance of understanding the primary function of an object, Cosmides and Tooby ask us to first imagine whether we could guess what the item in question is, if we were given a list of arbitrary features or properties. By way of illustrating this point, imagine trying to guess what an appliance is based on the following features: has metal parts, weighs seven pounds, and has a cord coming out of it. A list of features like this—Cosmides and Tooby emphasize—would not enable us to know what the focal object is, and more importantly, what it is for (that it is a toaster and that its function is toasting bread). Nor would the above three features help us understand how the various sub-components of the toaster enable it to perform its main function. After all, based on the three features, the item in question could be an indefinite number of things, including a speaker for a stereo, a rechargeable battery pack, an electric hand mixer, a small vacuum, a telephone, a laptop, or perhaps a table lamp. Furthermore, a listing of “functionally arbitrary” properties—that the toaster has a certain weight and could therefore be “heavy enough to use as a paperweight”—would merely lead us astray in trying to guess what the object is and what it was designed to do (Cosmides and Tooby, 1994a). Focusing on the main function of an object is critical to evolutionary analysis. As put by Cos- mides, Tooby and colleagues, “function provides a privileged frame of reference” (Klein et al. , 2002: 309). Taking this privileged frame means that any list of other properties or functions are merely characterized as “byproducts” that are not “co-equal” with the primary function Downloaded from https://academic.oup.com/icc/advance-article/doi/10.1093/icc/dtad057/7332166 by University of Oxford user on 31 October 2023 4 T. Felin and S. Kauffman (Klein et al. , 2002: 309). Multiple uses or functions are seen as irrelevant to understanding evo- lutionary fitness (as we discuss below, they are seen as rare exceptions: Dennett, 1995: 216). Furthermore, it is only by focusing on the main, historical function that we can uncover the pre- cise features and subcomponents that enable it to perform what it was designed to do (in the context of toasting bread: the heating elements, the slots, outer casing, and other components serve sub-functions to enable the main function). Importantly, the “reverse engineering perspective...sees all living things as efficiently composed of parts with functions” (Dennett, 2017: 54). This idea of hierarchically isolating functions and showing their optimality is central to evolutionary theorizing and associated functional analy- sis. 4 To briefly illustrate this point more broadly—beyond the metaphor of a toaster—Neander highlights how the analysis of “the human digestive or circulatory system, for example, proceeds by way of a decomposition of the system into its functionally individuated parts...And then a description of the proper function of each part is given, that being the contribution of each part to the functioning of the whole, when the system is functioning properly” (Neander, 1991: 467; also see, 2017). So far we have quoted extensively from existing sources—the work of Cosmides, Tooby, Dennett, and others—to assure the reader that we are not creating a caricature of the exist- ing literature. The bottom-line is that the extant literature puts a strong emphasis on isolating and establishing the primary function of any given trait, feature, or object—and a strong empha- sis on understanding function by reverse engineering. These are the central building blocks of adaptationist evolutionary theory. However, we offer an alternative conceptualization. To foreshadow our alternative, a central problem with the existing approach is that any given trait, feature, or object is reduced to one, isolable and historical function. This might make sense— for some types of functions, situations, and explanations. However, isolating a singular function presumes a type of essentialism that does not account for emergent function and associated evo- lutionary problem solving and innovation. Thus, we argue next that functional excess in fact is pervasive in nature—a premise that has wide-ranging consequences for how we understand evo- lution and emergent innovation. We might note that the idea of functional excess—while not his precise framing or point—was foreshadowed by William James. He argued: Many objects of daily use—as paper, ink, butter, overcoat—have properties of such constant unwavering importance, and have such stereotyped names, that we end by believing that to conceive them in those ways is to conceive them in the only true way. Those are no truer ways of conceiving them than any others; there are only more frequently serviceable ways to us (1890: 224). And James’s key punch line—particularly as it relates to our argument about function—is that: “there is no property absolutely essential to one thing. The same property which figures as the essence of a thing on one occasion becomes a very inessential feature upon another” (James, 1890: 222, emphasis in the original). We broadly concur with this logic. But we extend it into the domain of evolution and innovation more broadly. 3. The alternative—from functional essentialism to functional excess We argue that functional excess rather the functional essentialism characterizes nature. This sug- gests the need to move from retrospective functional essentialism to a perspective that recognizes the ubiquity of functional excess and the forward-looking opportunities that this enables for experimentation and innovation in evolution. We discuss this first by offering an extension of the 4 As recently summarized by Dennett, “reverse engineering is methodologically committed to optimality consid- erations. ‘What is—or was—this feature good for?’ is always on the tip of the tongue; without it, reverse engineering dissolves into bafflement” (2017: 122). Our emphasis is not on optimality (especially past optimality) of a given object for a particular function, but rather on the ubiquity of functional excess and how an object might solve an evolutionary novel problem encountered by an organism. This distinction is critical, and a point to which we return throughout the manuscript. Downloaded from https://academic.oup.com/icc/advance-article/doi/10.1093/icc/dtad057/7332166 by University of Oxford user on 31 October 2023 Disruptive evolution 5 toaster metaphor. Thereafter, we discuss how functional excess is harnessed, made salient, and utilized in evolutionary settings—including both biological and economic ones. To illustrate functional excess, imagine a human stranded on a remote desert island, with only a toaster. While this might at first appear to be a strange example, it offers a useful thought experiment for discussing function and evolution. So, in the context of a desert island, what could a toaster be useful for? How might it enable survival? Notice how many of the seemingly arbitrary features of the toaster—features that previously were labeled irrelevant byproducts (like, say, the toaster’s weight or its shiny outer casing)—now take on completely new meaning, salience, and primacy. To first state the obvious, on the desert island the main function of the toaster—the thing it was designed and selected for, toasting bread (Cosmides and Tooby, 1994a)—is useless. And even the specially designed parts of the toaster have no purpose relative to their intended and selected function, as the main function of the toaster is co-dependent on functions related to having electricity. But the desert island—representing a new selective environment, problem, or situation—does not render the toaster useless. To the contrary, the situation and problem now creates salience for a whole host of new functions and possibilities. Varied features and parts of the toaster can readily be used to solve any number of critical problems to enhance survival. For example, the shiny outer metal casing of the toaster could be used as a much-needed reflection device to alert potential rescuers in planes or boats in the distance. Or the toaster with its cord could be used as a flail, perhaps to hunt with or to defend against attacking predators. One might also detach and use many of the sub-components and sub-features of the toaster in new ways. For example, one of the aluminum panels of the outer case could be removed and fashioned into a concave shape to concentrate sunlight to start a fire. The internal screws, coils, and other metal elements— which support the now useless main function of toasting bread—could be fastened to a stick and repurposed into sharp edges or implements for hunting or fishing. Or the toaster’s crumb tray could be used as a griddle to grill fish or toast coconut. Even the crumbs from the tray could be used as, say, bait to attract rodents, birds, or fish. Our central point here is that the list of possible uses—for any object—is indefinite. Any effort to try to list all possible uses for a given object would require a delineation of (a) all situations and problems, (b) all objects with which one might interact or combine the focal object in ques- tion, and (c) all organisms and organism-specific capacities for using that object (size, dexterity, strength, cognitive capacity, etc.). The resultant problem–object–organism matrix is indefinite, leading to a combinatorial explosion of functional possibilities. This combinatorial matrix could further be expanded by the fact that what constitutes a relevant object or feature is also not knowable in advance (for example, the “curve” of the toaster could have a novel use, depending on the situation). Function is situation and problem-dependent. 5 The fact that the outer casing of the toaster is shiny is irrelevant to the primary function of toasting bread. But this particu- lar feature becomes highly relevant—even critical—in the context of needing to signal distant rescuers. Function is therefore the result of the nexus of interactional factors related to the sit- uation or problem at hand, as well as the organism in question. These interactional possibilities are impossible to pre-state ahead of time. Undoubtedly, novel functionality might also emerge randomly, in the colliding, mixing, and serendipitous interaction of objects and “parts” that find novel uses (as is the case with the swim bladder—see Felin et al. , 2014: 275–276). But functional excess also provides a reservoir of dormant raw materials and possibilities that organisms can tap into to solve evolutionary problems in a forward-looking fashion. Notice that the recogni- tion of this type of dormant excess or “poly”functionality does not deny the fact that a particular trait or object might have had a historically critical function that enabled survival. However, it is also important to recognize dormant functional excess, as this provides the raw material for subsequent evolutionary innovation. 5 It is important to note that function is not just problem-dependent (as well as organism-dependent), but also availability -dependent. Certain objects are more likely to be present in particular situations and thus available to poten- tially be used in new ways. While this is an important point, our focus is on an adjacent one, namely, that environments “teem” with many objects that have indefinite functions which might be available and useful for solving particular problems (for further discussion, see Felin et al. , 2023). Downloaded from https://academic.oup.com/icc/advance-article/doi/10.1093/icc/dtad057/7332166 by University of Oxford user on 31 October 2023 6 T. Felin and S. Kauffman Our focus on functional excess builds on a different biological tradition, one that has received far less attention (compared to extant work focused on reverse engineering). This tradition rec- ognizes organism-specificity as well as the situation-contingent nature of function. This approach to function is partly captured by the biologist Jakob von Uexküll who—in his book Theoretical Biology (Uexküll, 1926)—makes a distinction between “leading” and “accompanying” proper- ties of an object, trait, or feature. As he argues, what constitutes a leading property of an object (that is, its main function, presently) versus an accompanying property is inherently situation- dependent. As he puts it, “leading and accompanying properties change with the change of function” (Uexküll, 1926: 108). Objects are constituted by what Uexküll calls a vast “residue” of accompanying properties (Uexküll, 1926: 109). It is this excess or residue of function that, we argue, can be harnessed in varied ways to solve evolutionary problems. Now, we need to point out that while evolutionary scholars have emphasized the histori- cally primary function of traits and objects, there are some instances where excess functionality is recognized. These instances are extremely instructive, as they open the door for rethinking how function is treated in evolutionary arguments more broadly. To illustrate, in his discussion of evolutionary functional analysis, Dennett (1995: 212–234) points out how an early home computer—orders of magnitude more powerful than anything Turing had access to—might pri- marily be used for something simple like word processing. Dennett asks, “why is all this excess functionality found here?” After all, home computers clearly offer far more information process- ing capacity than what is required for simple word processing. Dennett argues that the word processor is “locked into menial service”—because it is “performing a minuscule fraction of the tasks it could be harnessed to perform” (emphasis in the original). Notice that Dennett empha- sizes that the home computer is a rare example of excess functionality, as nature tends to design optimally and not waste resources (Dennett, 1995: 212–220; 232–234). 6 Our argument is directly contrary to Dennett’s. We claim that any object—not just Den- nett’s word processor—is performing a miniscule fraction of the tasks it could be harnessed to perform. Excess functionality is ubiquitous rather than a rare exception. In a sense, every object is locked into menial service, given indefinite, dormant possibilities. While we are sur- rounded by this excess functionality, it lies dormant and hidden. Excess functionality is dormant because it is presently not recognized or needed—or the right situation, application, or problem has yet to present itself. Without an obvious problem, excess functionality is not easy to spot. However, the presentation of a specific problem or situation—like “a-human-stuck-on-a-desert- island-trying-to-survive-with-a-toaster”—informally highlights how salience for novel functions can emerge. A natural response to this line of argument might be that it is all too easy and convenient to come up with seemingly ad hoc examples of function: artificial situations or problems where some object—or object sub-feature—turns out to be useful or is used in a novel way. After all, extant evolutionary arguments insist that what something is “normally” and historically used for should be the focus of evolutionary analysis. As articulated by Millikan, “associated with each of the proper functions that an organ or system has is a ‘normal explanation’ for the performance of this function, which tells how that organ or system historically managed to perform that function” (Millikan, 1993: 243, emphasis added , for an extensive discussion of “normal-proper” function, see Neander, 2017). Certainly there is nothing normal or proper about using a toaster as a signaling device to alert rescuers. But this is precisely our point. Function cannot be pre-specified without a context, problem or organism. Importantly, an organism facing a particular problem is not concerned with the proper or historical function of an object. It is merely interested in whether an object at hand helps solve the problem. For a cockroach or an ant, a toaster can serve the function of being a hiding place from a human trying to kill it. Here, the problem and organism’s particular physiology (size) are amongst the central variables enabling the toaster’s function of “useful-to-hide-in/under.” Situation-dependent and organism-specific factors are critical for understanding function. This 6 Dennett argues that evolution is frugal and does not, most of the time, allow for excess function. But he argues that the chip in the computer is a rare exception and explainable due to its low price, that is, because chips are the “best, cheapest way to solve the problem” (Dennett, 1995: 216). Downloaded from https://academic.oup.com/icc/advance-article/doi/10.1093/icc/dtad057/7332166 by University of Oxford user on 31 October 2023 Disruptive evolution 7 logic is aptly captured by Uexküll’s idea of the species-specific surroundings or Umwelt (for fur- ther discussion, see Felin and Kauffman, 2021). Functions are given by the nature of the organism and the set of problems it encounters. Thus, “the stone that a beetle climbs over is merely a beetle-path, and does not in any way belong to the science of mineralogy” (Uexküll, 1926: 130). Given our radically expanded view of function, one might rightly ask: “is all functionality just in the eye of the beholder?” (McKay and Dennett, 2009: 495). Our argument is not that “anything goes.” Rather, our argument is pragmatic: a function is a function if it solves the prob- lem at hand. And whether something constitutes a function is ultimately a testable proposition, something that can be experimentally tested. However, there is no exhaustive way of delineat- ing functions. Despite this, evolutionary theorists have made themselves the all-seeing arbiters of what is and is not a proper function for a given object (cf. Felin et al. , 2017. Again, we recog- nize that establishing historical function might be important for some purposes. But historical function does not foreclose forward-looking possibilities and adjacencies (cf. Newman, 2023). In singularly fixating on the past, evolutionary scholars have not recognized the polyfunctional possibilities of objects and their potential to serve as dormant raw material for forward-looking problem solving, experimentation, and evolutionary innovation. Possible functions—what some- thing could be used for (by a particular organism in a specific situation)—represent the dormant raw material for solving evolutionary problems in forward-looking fashion. But this harnessing of dormant function is not automatic. 4. Rationality and the interactional anatomy of function Before proceeding to our arguments about how organisms can harness functional excess in vari- ous evolutionary contexts, we need to make two additional points related to functional analysis in an evolutionary context. First, we briefly discuss an underlying assumption about rationality that is presumed when considering evolutionary function (specifically, the supposedly computa- tional and statistical nature of rationality), and second, we highlight the necessarily interactional, physiological, and embodied nature of function. First, evolutionary theories treat the mind (and brain) as an evolved, computational, information-processing device (for a recent review, see Mastrogiorgio et al. , 2022; also see Anderson, 2010). This is aptly captured by Duchaine et al. who argue that “the human brain is a set of computational machines, each of which was designed by natural selection to solve adaptive problems faced by our hunter–gatherer ancestors” (Duchaine et al. , 2001: 225). As further argued by Cosmides and Tooby, “the brain was designed by natural selection to be a computer...composed of programs that process information” (Cosmides and Tooby, 2013: 203–204). And to understand the brain, we need “to reverse engineer [it]: to dissect its com- putational architecture into functionally isolable information-processing units” (Duchaine et al. , 2001: 225). This logic has also been applied from the evolutionary context to an economic under- standing of the mind and rationality (e.g., Cosmides and Tooby, 1994b), where humans are seen as Bayesians and intuitive statisticians (e.g., Gigerenzer, 2020). The brain then is seen as mas- sively modular, where each module has a specific function that was selected to solve a specific, frequently occurring evolutionary problem of the past. The information processing modules of the brain are seen as a highly specific statistical toolkit—humans are seen as intuitive statisticians (e.g., see Gigerenzer, 1992: 336). We build on a radically different view of rationality and the mind, particularly as it relates to function and evolutionary problem solving. We see the mind as a generative organ (cf. Chater et al. , 2018; Felin and Koenderink, 2022). That is, organism-specific problem solving and ratio- nality is not a computational or statistical task, rather it is forward-looking and attuned to novel possibilities (Felin et al. , 2014; Bongard and Levin, 2021). Our focus on the mind also sug- gests a completely different view of the environment. The environment is not represented by organisms, per se. The environment is too broad of a catchall term. Organisms are not engaged in information processing or computation. Rather, organisms engage with specific objects in their species-specific surroundings. Thus, we prefer to focus on how particular objects “light up”—present themselves and become salient—as organisms seek to solve problems. In this sense, Downloaded from https://academic.oup.com/icc/advance-article/doi/10.1093/icc/dtad057/7332166 by University of Oxford user on 31 October 2023 8 T. Felin and S. Kauffman rationality is also embodied and embedded in the functional possibilities afforded by objects in an organism’s surroundings. Accounting for these types of factors in any type of formal, statistical, or computational sense is not possible (Roli et al. , 2022). The specific mechanisms behind this are discussed in the next section. Our second point, building on the point above, is that functional problem solving is embod- ied and interactional— or put differently, not merely “in the head.” Functions and evolutionary problem solving necessarily implicate the interaction of both physiological features and objects that one might be surrounded by. Thus, we do not see how it is possible to assume the “ crisp anatomical isolation of functions (the kidney is entirely distinct from the heart; nerves and blood vessels are separate conduits strung through the body, etc.)” (Dennett, 1995: 216). Certainly we recognize that organs such as, say, kidneys or hearts, have functions that are critical for survival. That said, every organ also reciprocally depends on other organs, and the organism as a whole, for its function and survival (for example, the heart is tied to the circulatory system, autonomic system, and the organism itself). If we only focus on the mind, we leave much unexplained. Importantly, the idea of functional isolation is also highly problematic if we try to extend this logic not just to generative organs like the mind, but also to, say, the hands. It is hard to say what the primary function of an anatomical structure or organ like the hands might be. After all, hands can be used for waving or signaling, gesturing, pointing, grasping, counting, measuring, clawing, touching, holding, digging, poking, probing, flicking, clapping, rubbing, snapping, grabbing, squeezing, pinching, tying, lifting, pushing, writing, and throwing. Any of these functions—in combination with indefinite objects an organism might be surrounded by—can readily be linked with problem solving and survival, especially when coupled with other anatomical features (like the human mind). As summarized by Aristotle in his book On the Parts of Animals , “the hand is not to be looked on as one organ but as many; for it is, as it were, an instrument for further instruments”—or as put in a different translation, “the hand is the tool of tools” (Aristotle, 1984, circa 350 bce: Book I, Part XI). Animals of course engage in varied types of problem solving and innovation using tools as well (Köhler, 1925; Sanz et al. , 2013; Griffin and Guez, 2014; Fragaszy and Mangalam, 2018; Gruber et al. , 2019). And humans are particularly adept at “secondary tool use,” where tools and multistep processes and jury-rigging are used to solve problems in novel ways (Beck, 1980; Osiurak et al. , 2020; cf. Newson and Richerson, 2021; Roli et al. , 2022). 7 In most situations of evolutionary importance, it is quite evident that no trait, feature, or object works in isolation. Function is necessarily linked not only to other physiological organs or characteristics but also to objects in the environment. Making and using a spear is not just a cognitive activity but also implicates hands, arms, and the rest of the body, and other objects in the surrounding environment. In the human context, these types of mind–body–object interactions are the basis of indefinite functional possibilities. The human mind–hand interaction is an object– object or organ–organ interaction that has been particularly critical and generative in evolution (and in an important sense, what sets humans apart from other animals: e.g., Ambrose, 2001; also see Wilson, 1998). When it comes to problem solving, minds interact with hands, and hands are enabled by arms, bodies, and objects in the surrounding environment. An evolutionarily vital activity—like foraging or hunting—implicates all of these and other organs (the ability to run or move, the ability to smell). And interactional anatomical linkages in function can be vastly multiplied: their possible forms are combinatorially explosive, particularly in interaction with the set of objects one might find in the environment. In all, the central premise of the “relative rarity” of multiple functions—as well as the demand for the “strict isolation of functional elements” (Dennett, 1995: 212–220)—is highly problematic from our perspective. 5. Evolution and harnessing functional excess Given that functional excess is ubiquitous but dormant, what is the specific mechanism through which this excess might be harnessed? How specifically does a new use or function for an object— or set of objects—“light up” and become salient? While we have highlighted how a particular 7 Along with the work of biologist Jakob von Uexküll (Uexküll, 2010), the “embodied” nature of living in the world comes through in the work of phenomenologists like Merleau-Ponty (1945). Downloaded from https://academic.oup.com/icc/advance-article/doi/10.1093/icc/dtad057/7332166 by University of Oxford user on 31 October 2023 Disruptive evolution 9 problem or situation might create salience for a new use (using the toaster-island example), next we discuss how directed experimentation—and “protoscience” more generally—plays a critical role. Thereafter, we discuss the implications of these arguments for not just biological settings but also economic ones. To make our arguments concrete, consider a particularly important evolutionary activity like search. Search—including activities such as foraging or hunting—is perhaps amongst the most fundamental activities in evolution. As noted by Hills et al, “search is a ubiquitous property of life” (Hills et al. , 2015: 46). Organisms search for various things such as mating partners, shelter, nesting locations, and food. And economic actors search for value. To represent se