The animal sensorimotor organization: a challenge for the environmental complexity thesis

Godfrey-Smith’s environmental complexity thesis (ECT) is most often applied to multicellular animals and the complexity of their macroscopic environments to explain how cognition evolved. We think that the ECT may be less suited to explain the origins of the animal bodily organization, including this organization’s potentiality for dealing with complex macroscopic environments. We argue that acquiring the fundamental sensorimotor features of the animal body may be better explained as a consequence of dealing with internal bodily—rather than environmental complexity. To press and elucidate this option, we develop the notion of an animal sensorimotor organization (ASMO) that derives from an internal coordination account for the evolution of early nervous systems. The ASMO notion is a reply to the question how a collection of single cells can become integrated such that the resulting multicellular organization becomes sensitive to and can manipulate macroscopic features of both the animal body and its environment. In this account, epithelial contractile tissues play the central role in the organization behind complex animal bodies. In this paper, we relate the ASMO concept to recent work on epithelia, which provides empirical evidence that supports central assumptions behind the ASMO notion. Second, we discuss to what extent the notion applies to basic animal architectures, exemplified by sponges and jellyfish. We conclude that the features exhibited by the ASMO are plausibly explained by internal constraints acting on and within this multicellular organization, providing a challenge for the role the ECT plays in this context.

Keywords: Environmental complexity thesis, Early nervous systems, Skin brain thesis, Sensorimotor system, Epithelia, Multicellular organization

In 1996 Peter Godfrey-Smith published his book Complexity and the Function of Mind in Nature. Here, he first put forward his influential environmental complexity thesis (ECT) as a general principle about the evolution of cognition: “The function of cognition is to enable the agent to deal with environmental complexity.” This succinct phrase reflects an insightful and wonderfully intricate set of ideas and reflections on the ways organisms derive advantages by using cognitive capacities to coordinate their behavior with the state of the environment (Godfrey-Smith 1996a, b, 2002). Here we will focus on, and challenge, the idea that the ECT provides a sufficient account for the evolutionary origins of animal cognition. We treat the ECT as an externalist proposal that prioritizes environmental heterogeneity as the key feature for the evolution of cognition. This does not imply that we interpret Godfrey-Smith’s position as a form of simple asymmetric externalism (properties of the environment fully explain the properties of the organic system, but not the other way around) a position that he criticizes himself. Godfrey-Smith’s position combines internalist and externalist elements, and the same goes for our own view defended below. The difference with Godfrey-Smith concerns the emphasis and relative importance that we respectively place on these organismal and environmental aspects. Treating his position as an externalist one merely reflects that for the ECT the environment plays a more prominent role than the agent.

Our focus will be on the notions of agent and environment that both feature in the ECT. These notions come together as an agent implies some environmental context in or on which it operates, an issue that we will address below. While Godfrey-Smith (2002, in press) discusses the possibility of ‘proto-cognition’ in bacteria, plants and less complex animals, the default target of the ECT are complex animals and the environment they inhibit. Complex animals—in particular humans—are after all the standard targets when we think of cognition and the mind. In these cases, the ECT applies to animal—and thus multicellular—agents acting on and reacting to a macroscopic environment composed of recognizable objects (trees, grass, predators etc.) and various media available for locomotion (e.g. water, soil and various surfaces). We claim that neither animals—being multicellular organizations—nor the macroscopic environment in which they act can be taken as a self-evident starting-point for the evolution of cognition.

O’Malley (2014) recently sketched the far-reaching philosophical implications of current microbiology and argued that life should be considered microbial by default, while multicellular macrobial forms are specializations that only occur in a limited number of lineages. The implication is that animals and other multicellulars are special cases the occurrence of which requires additional explanations on top of a more basic cellular organization. Even if you find this claim too strong, the fact remains that animals consist of complex and peculiar multicellular organizations that in their modern form only became visible with the Cambrian Explosion, about 542 Ma, in contrast to unicellular life that has been ongoing since about 3.5 Ga. Trestman (2013) used the phrase complex active bodies to refer to these new entities that he describes with a cluster of related properties: “(1) articulated and differentiated appendages; (2) many degrees of freedom of controlled motion; (3) distal senses (e.g., “true” eyes); (4) anatomical capability for active, distal-sense-guided mobility (fins, legs, jet propulsion, etc.); and (5) anatomical capability for active object manipulation (e.g., chelipeds, hands, tentacles, mouth-parts with fine-motor control)” (p. 81). Such properties are exactly those that allow animals to act on the macroscopic environment of which we—as humans—have firsthand experience. It is an important evolutionary question how and why these modern animal features arose.

In contrast to work on the Cambrian Explosion of complex animal bodies (Marshall 2006), much less is written, or understood, about the prequel of this key event (Keijzer 2015). Trestman rightly stresses that ‘complex active bodies’ require additional, cognitive features to make them operative. He proposes the initial presence of “a cognitive toolkit for pickup, tracking, and use (for control of behavior) of certain information from the environment” (2013, p. 89), which would be realized by some form of nervous system. However, where did such a convenient neural tissue organization and its characteristics come from in the first place? Trestman acknowledges this problem when he writes that these requirements “do not simply follow from selection pressure due to predation or competition, but requires some (unknown) set of preadaptations in neural control architecture, which must be to some extent fortuitous—there is no foresight in evolution” (p. 91). If this claim about preadaptations is right, and we think it is, this spells a difficulty for the ECT at its very starting point: Multicellular agents but also the environments in which they act are themselves in need of an explanation (e.g. Arnellos and Moreno 2015).

The evolutionary transition from a collection of cells to active animals has so far remained outside of the scope of the ECT. However, this transition is not clear cut at all. Bonner (2000) mentions a minimum of 13 independent transitions to multicellularity, others discuss an even higher count of 25 (O’Malley et al. 2013), the animal case being only one of these. None of the other cases did turn into anything even remotely resembling the free-moving life style that is typical for complex animals. Thus, the evolution of an animal multicellular organization, in particular macroscopic, freely moving agents, cannot be taken as a self-evident, evolutionary next step that is to be expected when multicellular lifeforms arise.

In this paper, we address the question how and why this evolutionary transition might have taken place and the role of early nervous systems therein. Our discussion comes close to some of Godfrey-Smith’s own recent work (Godfrey-Smith, in press; Jékely et al. 2015a), which recognizes a distinction between Internal Coordination (IC) and Input–Output (IO) views on the role of (early) nervous systems. An IC view highlights the need to “induce and coordinate activity internal to large multicellular organizations” (Jékely et al. 2015a, p. 1) and can be contrasted to a more environmentally oriented IO view (ibid.) that is most easily aligned with the ECT. While both views can, and must, be combined, their relative importance is an issue of discussion. Here, we will press the claim that IC issues provide the key to explaining the transition to the modern animal organization.

The main problem we explore concerns the features that are required to build and constitute a multicellular organization that is sensitive to its own and its environment’s macroscopic organization. There is an old metaphor for the mind where the world acts as a stamp that imposes its structure on our memory, envisioned as a wax tablet. This metaphor assumes a structure, a wax tablet, which is sensitive to the stamp and retains the information it imposes once the stamp is removed. While sensitivity to environmental features at a biomolecular and unicellular scale is essential to maintain any living organization, in animals with active bodies an organization is present that has such sensitivity at the multicellular level. True (image forming) eyes are a good example. Here the spatial visual structure is somehow taken up by the animal, in a way that is not organized within any individual cell, but spread out across several cells where the position of these cells with respect to one another becomes central. In this case, the individual cells are no longer the single level at which the organism interacts with ‘the’ environment but the multicellular organization brings in its own additional organization allowing the environment to expand into the macroscopic configurations that we ourselves inhabit.

Where does this multicellular organization come from and how does it function? Recently, this problem has been targeted by an IC proposal concerning early nervous systems (Keijzer et al. 2013). This ‘skin brain thesis’ proposes that the coordination of contractile epithelia for motility formed the key adaptive feature behind the earliest nervous systems that evolved, presumably long before the Cambrian Explosion (Erwin 2015; Wray 2015; but see also Budd 2015). In addition, Keijzer (2015) argued that such a spatially extended proto-neural and contractile system also provides the basic ingredients for a multicellular sensing organization: Self-induced epithelial contraction combined with stress-induced cellular feedback can connect macroscopic patterns across a multicellular epithelium to specific events at a cellular level. The result can be envisioned as a small slug, consisting of connected contractile cells, that controls its overall body-shape in a reversible way by manipulating the contraction levels across its body in a way that is sensitive to both self- and externally induced changes in these contraction levels at a cellular level. A configuration like this could solve the question how macroscopic happenings that extend across a multicellular organization became able to ‘talk’ to, and in this way become coordinated with, intracellular events. Metaphorically, a contractile biomolecular organization constituting the now active ‘wax tablet’ has acquired a means that enables it to tell apart the different macroscopic patterns it actively stamps on.

As argued by Keijzer (2015), such a tissue organization provides a foundation for a sensorimotor organization that is specific for animals that use contractile tissue to generate motility. Keijzer referred to this as an animal sensorimotor organization or ASMO. An ASMO is claimed to be present in all animals with nervous systems and muscles, and it provides a source for the capacity to pick up and use macroscopic patterns of environmental information that became typical for Cambrian style motile animals. In this paper, we develop the ASMO notion in some detail by discussing its relation to epithelia. Epithelia constitute the basic and most typical tissue organization exhibited by animals (Tyler 2003; Leys and Riesgo 2012). Here, we discuss how an epithelial organization and its properties can be cast as the key enabling factor for the evolution of the ASMO and also how a focus on the epithelial organization integrates the ASMO more closely to development and physiology; all essential features of the organizational basis of complex animals (Arnellos and Moreno 2015, 2016).

Some may wonder whether animal bodies can be considered to be part of the cognitive domain. However, within embodied approaches to cognition, the notion of cognition is already applied in a wide sense (e.g. Barrett 2011; Chemero 2009) and the same goes for ongoing discussions on bacterial (Lyon 2015) and plant cognition (Calvo and Keijzer 2011). In this context, the foundations of the animal sensorimotor organization easily fulfils the criteria for a minimal reading of cognition.

Without addressing the ECT directly here, the ASMO notion sets up a challenge for the ECT nevertheless. While the ECT stresses environmental complexity as the main driving force towards complex forms of intelligent behavior by animals, the ASMO notion highlights the prior importance of the (internal) multicellular organization as a precondition for the macroscopic environment inhabited by animals to become accessible for these animals. The challenge for the ECT is whether the origins of the animal sensorimotor organization—on which the ECT depends—can also be explained in terms of environmental complexity or whether an IC account has the better cards here.

The paper has the following structure. In the section “The animal sensorimotor organization (ASMO)”, we give a short description of the ASMO notion. “Epithelia and their features” discusses the characteristics of epithelia, while in “Epithelia and the animal sensorimotor organization” we sketch how epithelia provide a very suitable foundation for the characteristics of the ASMO. In “Two model animal organizations: Porifera and Cnidaria”, the features of the ASMO are related to typical characteristics of Porifera (sponges) and Cnidaria (such as jellyfish). Finally, “Bodily complexity meets environmental complexity” provides a short discussion how the discussion on the ASMO sets up a challenge for the ECT.

The ASMO is a solution to a problem that is not always clearly perceived: Acquiring a biological organization that is sensitive to macroscopic environmental features, including the macroscopic features of the body itself. The problem derives from the transition of single cells capable of a sensorimotor interaction with their environment to a multicellular organization doing the ‘same’ but at a significantly larger scale. As long as we discuss this interaction only in terms of input and output—terms that can be described in many different ways—it is not obvious that there are major organizational changes involved in this switch (Keijzer 2015). One important reason seems to be the word ‘brain’ itself. The concept of a brain is often used like that of a material mind that can handle environmental descriptions in many different forms without needing a clear account how this tissue might achieve such a feat. While it is obvious that brains are actually capable of performing this task, there are two important questions here: (1) How do nervous systems function? (2) How could any particular tissue configuration have acquired this ability? Natural selection does not fulfil this exploratory burden on its own as it needs traits to work on (see for instance Arnellos and Moreno (2012) for a detailed discussion of the organizational complexity of protocells). The issue here is how various traits—like sensing and motion control—that are standardly used with respect to brains originated at a very early and basic level in animal tissues.

Multicellular organizations—most notably animals, plants and fungi—organize the way in which they act on and react to their communal environment in various ways. The focus here is on the organization that has evolved within the animal lineage and given rise to their ‘complex active bodies’ as mentioned above. The central idea behind the notion of an ASMO is that complex animal senses did not evolve by stacking individual sensory cells together to form static sensory arrays. Static arrays are implausible for early animals that were soft-bodied and deformable, in particular for the basic sense of touch. Therefore, instead of interpreting animal sensing as an input process, the ASMO notion claims that animal sensing derives from a slug-like organization of cells that are physically connected to one another and capable of contracting. By being physically connected, cells transmit the generated tensions across the whole slug-body. The generated changes in tension and shape across the slug-body can be used as a differentiated signal for the individual cells and their intra-cellular signaling, which in turn influences the slug-body again. Direct physical contact with the environment would influence the organism by impinging on the dispersal of tensions across the slug-body. This configuration distinguishes the ASMO from an input–output system with its clear conceptual distinction between an internal and an external domain. We propose that the ASMO constitutes an essential preadaptation that became the platform for complex sensors, effectors and nervous systems.

The notion of the ASMO derives from an account of the evolution of the first nervous systems (ibid.). This account centers on an idea of Pantin (1956) who stresses the need to coordinate the animal behavior machine: muscle. Motility induced by cilia can be used both by single cells and multicellular organisms, and is relatively easy to control (Jékely 2011; Elgeti and Gompper 2013). However, as argued by Pantin, the switch to contraction-based motility—which proved to be an essential step for the later evolution of complex, large animals—required more complex forms of coordination suggesting an important role for early nervous systems. Roughly, to squirt water from a balloon-like body, the animal will have to contract a significant portion of this body and control its body opening at the same time. This involves a fast way to synchronize and coordinate activity across sheets of physically connected cells. It is interesting to note here that nervous systems and muscle always come together in extant animals. Thus, in this internal coordination (IC) account, contractile and reversible changes in body-shape constitute the key phenomenon for early nervous systems. Most relevant for our present focus is that it enables the formulation of a specific animal sensorimotor organization that diverges from standard notions of input and output (Keijzer 2015).

The ASMO is a sensorimotor organization that is sensitive to the patterns of changes in stress across its contractile surface—referred to as the animal’s Pantin surface—in a way that reflects both self-induced and externally induced disturbances of these patterns. This organization, it is proposed, provides a way for macroscopic intercellular events (patterns across the Pantin surface) and (intra)cellular events to ‘talk’ to one another. Rather than being limited to localized reactions to environmental stimulation, this organization unifies a multicellular organization allowing it to deal with patterns at this bodily level. Rather than being limited to re-actions, this organization is intrinsically active and initiates its own changes of shape creating its own feedback of induced (reafferent) changes as well as exafferent changes that convey information about external features. This ASMO nicely fits the view of the sensorimotor organizations as influenced by work in ecological psychology (Chemero 2009; Turvey and Fonseca 2014) and sensorimotor theories (Hurley 2001; O’Regan and Noë 2001).

To develop the ASMO notion in more detail we combine it with current knowledge about epithelia and relate it to the basic animal configurations of Porifera and Cnidaria. For this discussion, we want to stress the relevant theoretical posits that will be discussed.

First, the ASMO relates to a particular organization, a way of putting a system together, that is not strictly tied to any specific historical scenario. It is a proposition about how individual animals might have functioned and potentially still function. For this purpose it builds on theorizing within embodied cognition that stresses the essential role for self-generated movement and the perceptual feedback this generates (e.g. Hurley 2001; O’Regan and Noë 2001; Barrett 2011). The evolutionary importance of these organizational features is plausibly cast as providing various organizations on which selection might have taken place (Calcott 2009).

Second, the ASMO is consistent with Input–Output (IO) views, such as the ECT. It focuses on different issues by highlighting the (internal) organismal organization on which much IO behavior depends (Jékely et al. 2015a). As said above, there can be some tension between the two however. For IO views like the ECT, accessing environmental information through senses is basic. The IC inspired ASMO notion suggests that a highly complex and specific multicellular organization provides a precondition for the ECT when this is applied to the macroscopic environments of modern animals. We will return to this possible implication in the final section.

Third, while the ASMO is formulated in the context of the evolution of early nervous systems, we do not focus here on the actual involvement of a full nervous system. What counts most is a spatially extended effector surface, electrical (or maybe chemical) signaling across that surface, and the various ways in which the patterns of tension differences across that surface might feed back to and influence the processes that initiate and are responsible for that patterning signaling. Nervous systems are thus not a precondition for the ASMO in this account, but rather a consequence of its presence and use.

To organize the discussion of the ASMO notion in this paper, we use five theoretically grounded conditions concerning the kind of living organization that can be expected to yield a basic ASMO. Thus we take the ASMO notion to subsume:

a multicellular body, constituting an ‘inner space’ or domain, which is differentiated from the body’s ‘outer space’ or environment.

The first four are central features of the ASMO notion as formulated originally by Keijzer (2015). The fifth is an addition that addresses the issue how contraction and ciliary motility can be related in a basic ASMO. While the ASMO eventually turned into the foundation of muscle-based motility, such motility itself derives from reversible changes in body-shape, whether used for motility or not. Below, we discuss how these posits line up nicely with various central features exhibited by epithelia, and allow us to extend the relevance of features that are already closely studied in developmental biology to the ASMO.

Epithelia are the key to the animal multicellular organization.1 Tyler (2003) argues that epithelial cells are the default eumetazoan cell state, while epithelia provide the central ingredient for developmental signaling as well as the actual construction material and developmental scaffolding for much of the animal body (Tyler 2003; Holland 2011). Some authors compare embryogenesis to origami folding: the flexible two-dimensional epithelial sheets bend and fold themselves in specific ways and in elaborate sequences, eventually producing the amazing three-dimensional structures that constitute the animal body (Holland 2011; Mao and Baum 2015). Epithelia, together with the cells that constitute them, are thus central to the animal organization and make it different in significant ways from non-metazoan multicellular organizations (Tyler 2003; Arnellos and Moreno 2016). In this section, we give a quick overview of some central features of epithelia with a general eye to their relevance for the evolution of the ASMO.

An epithelium is a two-dimensional sheet-like tissue that consists of polarized cells anchored in an underlying extra-cellular basal membrane (or lamina), which is part of the underlying extracellular matrix (ECM). The ECM is a continuum of collagen fibers and fibrils that integrates and reinforces the epithelium making it more resistant to tension and provides a stable setting for the epithelial cells (Timpl 1996). Epithelial cells have an outward-facing apical side with a primary cilium, a basal side connected to the basal membrane, while they are also connected to adjacent epithelial cells through a variety of junctions (Tyler 2003; Ingber 2006; Magie and Martindale 2008).

Epithelial cells have various types of junctions, which can differ between the various lineages. We discuss the ones that are most general and functionally relevant for the ASMO. Adherens junctions—mainly constituted by proteins of the cadherin family—provide strong physical connections between cells, tethering them together, while also making contact with the intracellular cytoskeleton, which is composed of actomyosin networks, microtubules and intermediate filaments (Magie and Martindale 2008; Engler et al. 2009). In addition to adherens junctions there are also integrin-based adhesion complexes—focal adhesions—positioned at the basal side of the cell, connecting it with the basal membrane. Focal adhesions are highly sensitive to forces imposed on the ECM (Engler et al. 2009; Geiger et al. 2009). A second central junction type consists of occluding junctions. These seal off the space between the epithelial cells for diffusing molecules (Skaer and Maddrell 1987; Magie and Martindale 2008), which creates molecular gradients that enable intercellular signaling and electrical differences (Cereijido et al. 2004). Occluding junctions allow epithelia to close of extracellular spaces from the general environment turning them into compartments or containers with a distinct molecular composition. Such compartments provide stabilized environments where various physiological processes can proceed without being disrupted by environmental changes.

Another central characteristic of epithelial cells is their capacity to move, to change shape and to respond to and actively impose mechanical stress. Epithelial cell use their actomyosin cytoskeletons to produce and resist such stress, which can travel through their adherens junctions and focal adhesions to both neighboring cells and the ECM (Wozniak and Chen 2009). This self-induced and transferred motility also drives the folding patterns during development. The mechanical forces generated by the cells’ cytoskeletons are essential for developmental processes such as proliferation, differentiation and spatial rearrangements (Ingber 2006; Chen 2008). Mechanical forces are now recognized as being involved in processes like wound healing (Röper 2013) and differentiation of stem cells (Nava et al. 2012), but they also influence gap junction coupling (Salameh and Dhein 2013) and the activity of ion channels (Sachs 2010). Thus mechanical forces are a key feature that influences epithelia at scales that range from the tissue level to the biochemical signaling within and between cells.

Epithelia further provide the context for intercellular signaling that together make epithelia the self-folding origami virtuosos capable of building the complex animal bodies we witness today. Epithelia provide a reliable scaffold to guide diffusing morphogens that initiate developmental processes like cell differentiation and various morphogenetic movements at specific places in ways that fit the global patterning process. Sensitivity to the spatial lay-out enables e.g. axial differentiation, cell-type specification and regional tissue patterning (e.g., Davidson and Erwin 2006; Newman 2012; Newman and Müller 2010). The result consists of developmental programs based on both intra- and intercellular signaling mechanisms that mobilize and drive morphogenesis.