On the prospects of basal cognition research becoming fully evolutionary: promising avenues and cautionary notes

Even countering strict multicellularism is a promising avenue to try to understand the evolution of multicellularity itself. Arias del Angel et al. (2017) have taken an Evo-Devo perspective on the aggregative multicellularity of myxobacteria. Their rationale is that investigating the diverse evolutionary origins and modes of multicellularity facilitates evolutionary comparative analyses. Such analyses can aid in distinguishing lineage-specific aspects of multicellular evolution (e.g., in animals or plants) from the generic factors and mechanisms underlying the transition to multicellularity across different groups of organisms (Arias del Angel et al., 2017). For scientists with an Evo-Devo mindset, the origin of animal multicellular development, for instance, is sought in the developmental trajectories and developmental potentialities of closely related lineages (Brunet et al., 2019; Dayel et al., 2011; Fairclough et al., 2010; Ruiz-Trillo et al., 2023).

Also, for basal cognition advocates destabilising and resisting strict neuronalism opens up new research avenues. Recent biological research focusing on the behaviour of various unicellular organisms such the ciliate Stentor coeruleus (Rajan et al., 2023) and the acellular slime mould Physarum polycephalum (Boisseau et al., 2016; Vogel & Dussutour, 2016; for discussion, see also Sims, 2024b) have provided evidence that habituation (a simple form of learning) does not require neuronal processing.7 Besides providing concrete reasons for cognitive science and comparative psychology to reconsider traditional implementation constraints that have been imposed upon learning across the board, these empirical discoveries have opened up new and important questions regarding different memory storage mechanisms. Whilst modification of synaptic strength may be one storage mechanism that straightforwardly requires neurons, cellular molecular storage does not (Gershman, 2023). Whether the molecular mechanisms underwriting memory storage in non-neuronal and neuronal organisms are shared is an open question, and–if we are on the right track–one which needs to be addressed along with questions about functional variation in memory and habituation across taxa. That said, resisting strict neuronalism has already hinted that it is worth its weight in salt.

Similarly to the case of multicellularity in Evo-Devo, understanding the evolutionary origins of neurons and nervous systems (e.g., nerve nets) requires investigating non-neuronal organisms within and outside the animal kingdom (Arendt, 2020, 2021). Doing so promises to provide a wealth of information upon which to make empirically based inferences about the economisation of organismal organisation that early nerve nets provided for basal metazoans (Keijzer et al., 2013). Likewise, recent work has found neuropeptide homologs in choanoflagellates, thus pointing to the pre-metazoan onset of their signalling functions (Yañez-Guerra et al., 2022), a relevant piece of information for understanding how they later were co-opted in nervous system evolution.

In general, evolutionary-driven approaches are promising for transcending biases and steering clear of reifying extreme positions (i.e., both oligoextensionalism and panextensionalism). Uncritically subscribing to panextensionalism, the assumption that development or cognition pervades all organisms, could impede scientific hypothesis construction and testing and, more generally, obscure the intricate relationships between the processes and capacities subsumed under these broad umbrella terms. In particular, if cognition is stipulated as a feature of all organisms, past and present, the largest evolutionary puzzle then resides in the origin of life itself (which would also be identical to the root of cognition), rendering subsequent inquiries on the evolution of cognition as mere complexifying elaborations on a common theme.

With this we arrive to another important cautionary note. In both developmental and cognitive sciences, oligoextensionalism and panexensionalism represent opposing extreme positions: development and cognition occur only in some forms of life (clonal multicellular and neuronal respectively) vs. development and cognition occur in all forms of life. Although extreme positions can often ignite and arouse interest in a particular field of research—especially for newcomers to a field or in the early days of a research field—radical positions often turn on deploying definition-based approaches and, as such, disputes tend to take the form of recalcitrant exercises in redescription—enter the many-headed hydra. However, as empirical discoveries are made, extreme positions can sometimes give way to what might look like a ‘golden mean.’ As Stephen J. Gould writes of the eventual closing of the gap between eighteenth century preformationists and epigeneticists schools of thought regarding heredity and embryonic development:

Modern genetics is about as midway as it could be between the extreme formulations of the eighteenth century. The preformationists were right in asserting that some persistence is the only refuge from mysticism. But they were mistaken in postulating preformed structure, for we have discovered coded instructions. […] The epigeneticists, on the other hand, were correct in insisting that the visual appearance of development is no mere illusion. (Gould, 1977, p. 18)

Our point is not to suggest that something like a golden mean should be the aim for oligoextensionalism or panextensionalism with respect to cognition and/or development8; rather, we would like to argue that deciding between these extremes is a matter of carrying out observations, formulating testable hypotheses, performing experiments (behavioural, structural, and molecular analysis), and letting the results of such experiments help scientists determine which organisms engage in which forms of cognitively driven behaviour and/or developmental processes (if at all) and how those specific forms might share commonalities or differ from those exhibited by other taxa. It was a similar kind of slow yet, as we now recognise, fruitful empirical process that led to the golden mean between preformationism and epigenesis which contemporary genetic research represents. This kind of process is core to a fully evolutionary approach. Similarly, we may expect that the polarising tension between oligoextensionalism and panextensionalism with respect to cognition will likely be mediated by taking a dialectical evolutionary-driven approach. Such an approach views working definitions as starting points that can be adjusted as needed with new empirical discoveries, rather than as fixed endpoints requiring theoretical defence.9 Philosophical adherence to a priori, fixed conceptions of cognition, in both oligoextensionalist and panextensionalist camps, can hinder the formulation of hypotheses that diverge from these conceptions or obscure evidence that contradicts them. Basal cognition researchers would do well to remember this.

In fact, as philosophers of Evo-Devo have judiciously argued, it is often more productive to move beyond offering univocal definitions of scientific concepts and instead characterise the explanatory agenda associated with them. This involves identifying the specific questions being addressed by the relevant epistemic communities, delineating the research horizon of salient problems they are confronting, and clarifying the success criteria for their resolution. For instance, Brigandt and Love (2012) explored this approach in their discussion of ‘evolutionary novelty,’ and Nuño de la Rosa (2017) applied similar reasoning to the concept of ‘evolvability.’ The key insight from these analyses is that conceptual diversity can actually stimulate scientific research. Accordingly, basal cognition research might benefit from embracing different theoretical perspectives on what ‘cognition’ is that could foster interdisciplinary collaborations in addressing its core agenda: understanding the evolution of specific cognitive capacities (e.g., memory, learning, decision-making, and anticipation) and their mechanistic underpinnings across the tree of life.

In the next two sections, we delve deeper into the promising avenues and cautionary notes that sound phylogenetic thinking, another stepping stone en route to a fully evolutionary orientation, could bring to future research on basal cognition.10

In fields thoroughly grounded in evolutionary considerations, like Evo-Devo, multiple ‘styles of thinking’ are employed to study evolution. These include, in addition to the causal-mechanistic pursuits discussed in Sect. 2, the well-established ‘population thinking,’ which focuses on uncovering the evolutionary forces driving changes in trait frequencies across generations, as well as ‘tree thinking,’ essential for launching comparisons across species and mapping evolutionary relations, and ‘homology thinking,’ crucial for comprehending how organismal traits are individuated during development and how they can exhibit quasi-independence in their evolutionary trajectories (Wagner, 2016). Explaining the wide range of characters across the tree of life, including ‘cognitive characters’ (for discussion, see Figdor, 2022, 2024a), requires understanding various processes that have given rise to diversification of extant life and properly contextualizing them within their phylogenetic history. And for this, distinguishing between homologies and homoplasies is of paramount importance.11

By comparing lineages and organisms from different species, evolutionary biologists bring together tree thinking and homology thinking and aim to uncover ancestral relationships that underlie the shared identities of certain characters (i.e., homologies), and this is useful, say, in tipping the balance when gauging competing models of trait evolution in particular phylogenetic sequences. However, also cases of convergent evolution are epistemically significant in these comparisons, as they strengthen and help confirm hypotheses about adaptation (Currie, 2013), to give an example. Finding similar phenotypes in different organisms may be explained by retention from common ancestry (homology), but a careful phylogenetic appraisal may instead reveal that they are independently derived, due to convergence or parallel evolution, or that they experienced reversal to a plesiomorphic state. Such examples of homoplasy present opportunities for Evo-Devoists to discover potential underlying developmental mechanisms and components that may have been redeployed to produce the prima facie ‘same’ phenotypes (Wake et al., 2011). In this vein, Evo-Devoists have come to the realization that the properties, dynamics, and organizational patterns of development make certain analogous structures more likely to evolve (Minelli, 2019).

Drawing inspiration from Evo-Devo, a set of promising avenues for the field of basal cognition could be carved out by embracing ‘tree thinking’ and ‘homology thinking’ and benefiting from the noetic and explanatory resources they afford for their evolutionary investigations (Hall, 2012; Wagner, 2016). Along these lines, basal cognition research would benefit from differentiating between homologies and homoplasies when investigating cognitive traits and cognitive capacities. Discerning independent evolution and marking it off from bona fide uninterrupted (homologous) continuities, which cannot be presumed without the right kind of phylogenetic evidence, is epistemically decisive for avoiding ill guided inferences and unassured extrapolations.

The panextensionalist framing of basal cognition research that we explored in the preceding section unwittingly assumes that the similarities in the cognitive toolkit (see Sect. 2) are indicative of homology. For if the same cognitive capacities manifest in all extant life forms, from cyanobacteria and thermodesulfobacteriota bacteria to telonemids, spiny gulfweed and zombie fungi to hominids, it implies their presence in their most recent common ancestor—in this case, the putative LUCA, or ‘Last Universal Common Ancestor.’ However, merely uncovering similarity is insufficient to substantiate a particular evolutionary hypothesis concerning homology or homoplasy, for both insinuate similarities that require elucidation in the first place. Proper phylogenetic contextualization is a non-negotiable provision for putting these kinds of evolutionary inquiries on a productive track. This underscores a cautionary caveat: homology and homoplasy demand thorough and explicit assessment in basal cognition research, not mere surmises drawn from observing similar behaviours in organisms whose evolutionary kinships remain unclear (Figdor, 2024a, 2024b). With this, we do not want to rule out homology in (molecular, behavioural, etc.) traits that are related to the so-called basal toolkit of cognitive capacities, but rather point to the need to solidify the evidential apparatus for making such sweeping claims.

In extant organisms, we could witness similar capacities like memory, decision-making and anticipation, but it is entirely possible that some could be the product of common ancestry (e.g., conserved through uninterrupted negative selection towards variants that diminish their sizable contributions to organismal fitness) while others could have independently evolved in the different lineages giving rise to them. This is something that basal cognition advocates should evaluate and take up in their inquests to counter strict neuronalism. In this sense, probing the evolution of homoplasies can also be valuable for the basal cognition programme. For instance, multiple cases of convergent evolution have been uncovered in vertebrate nervous systems (for overviews, see Nishikawa, 2002; Strausfeld & Hirth, 2016) and some Evo-Devoists have adduced evidential support for the hypothesis that bilaterian central nervous systems evolved independently in chordates and protostomes (Holland et al., 2013; Martín-Durán et al., 2018). Correspondingly, convergent evolution needs to be investigated and cannot be ruled out for putative cognitive traits and even larger cognitive capacities in non-neural organisms from disparate lineages. For if, as basal cognition supporters claim, echoing Shettleworth's (2013) definition, cognition “is comprised of sensory and other information-processing mechanisms an organism has for becoming familiar with, valuing, and interacting productively with features of its environment in order to meet existential needs” (Lyon, 2020, p. 416), why couldn’t processes involved in acquiring, storing, and using information from diverse environments have evolved more than once in the history of life?

Moreover, there is no demerit whatsoever in a trait or process being the product of independent evolution. For example, fossil and molecular evidence suggest that megaphyllous leaves in euphyllophytes (ferns, gymnosperms and angiosperms) and microphyllous leaves in lycophytes (clubmosses, spikemosses and quillworts) may be homoplastic, having emerged several times in plant evolution (see, e.g., Corvez et al., 2012; Harrison et al., 2005). But this does not mean that these remarkable organs/structures linked to photosynthesis and transpiration found across euphyllophytes and lycophytes are not leaves or that we shouldn’t call them ‘leaves.’ Why cognition could only be called ‘cognition’ if it is tied to the lineages traditionally studied by cognitive science, as oligoextensionalists would have it? Or, as panextensionalists presume, if it is batch of capacities that all and every single organism instantiates? What are the arguments and the evidential basis to not even entertain the possibility of convergent evolution of cognitive traits and capacities throughout the tree of life?

Importantly for our foregoing comparisons, even when granting that (clonal multicellular) development evolved convergently in, say, plants and animals, we don’t have a problem to still describe their bundles of processes related to events like morphogenesis and differentiation as ‘development.’ Why should things be so different, borderline sui generis, when deliberating about cognition? We argue that not even countenancing, even if remotely, the possibility that certain cognitive processes and capacities might not be the outcome of an evolutionary singularity is antithetical to biological thinking. We don’t assert that plant development is not development because it is a product of independent evolution from animal development. Authors like Meyerowitz (2002) have argued that it is precisely because they are convergent that we can gain a lot of insight into what is truly general about development by comparing them (and at the same time, these comparisons accentuate the fact that certain processes are genuinely unique in both groups and cannot be generalized).12 This is another lesson, at the same time exciting and daunting, that is uncovered by our analysis.

In light of this backdrop, it would be advantageous for scientists and philosophers interested in basal cognition to give greater consideration to convergent evolution and homoplasies more generally in their research and especially in their emerging theoretical frameworks. For instance, in the recent double special issue in Philisophical Transactions of the Royal Society B on basal cognition, only one article has a section devoted to convergent evolution (Moroz et al., 2021), but it is restricted to treating neural systems in animals, and two other articles only cursorily mention it (Baluška & Mancuso, 2021; Boussard et al., 2021; but see Baluška & Mancuso, 2009). Likewise, none of the articles allude to homoplasy.13 This infrequent treatment, we believe, provides some evidence that this notion is not taken to be as important as homology—something that we think might be the by-product of their strong assumption of Darwinian continuity (Levin et al., 2021). Our suggestion for basal cognition research is not that it should simply be aware of homoplasies when countering strict neuronalism, but rather that it should countenance the importance of this evolutionary route and the frequency with which it happens and complement a more linear view of cognitive evolution with scenarios of parallel and convergent evolution of cognitive traits and capacities. A nice example in this direction is the recent discussion by Romanova and Moroz (2024) on convergent evolution of gravity sensors in unicellular protists and animals (e.g., Müller vesicles and statocysts), and the additional parallel evolution of gravisensory systems in basal metazoans. In the motley frame painted by the authors, gravitational sensitivity and the ensuing locomotory integrative systems cannot be understood through a view of linear cognitive evolution.

With this in mind, the evolution of cognitive traits and capacities, if treated with suitable phylogenetic tools and context, cannot be construed as a linear march of progress, as we will showcase with the next parallel.

By embracing phylogenetically-oriented thinking, besides being mindful of convergent and parallel evolution and taking appropriate steps for distinguishing between homologies and homoplasies across different (far-away and close) branches of the tree of life (see also Figdor, 2023, 2024b), the research undertaken under the basal cognition umbrella could profit from investigating phyletic evolution more closely (i.e., evolution within particular lineages). In this sense, it is important to stress that losses and gains of traits can come about, and the evolution of cognition should not be an exception on this regard.

As a general remark, over the course of phyletic evolution, one could attest relative complexification or simplification in the clades under study. When confronted with changing macroevolutionary patterns of biodiversity, biologists often appeal to “major transitions in evolution” (Szathmáry & Smith, 1995) and more strictly to “major transitions in individuality” (Michod, 1999) in an attempt to explain them. Roughly, such transitions identify hierarchical shifts in complexity that involve the emergence of higher-level units with novel (unit-dependent) features that arise from the coming together of lower-level entities. Such transitions have been interpreted to suggest that, in some lineages, there is an evolutionary trend for organisms to become more complex vertically (i.e., increasingly nested) and horizontally (i.e., possessing a larger number of different component parts). Primary examples are transitions from prokaryotic to eukaryotic life and from unicellular to clonal multicellular eukaryotes. In this regard, Evo-Devoists ask, for instance, if there is co-evolution of certain organelles (e.g., Woronin bodies of the Ascomycota phyla) and multicellular complexity (Jedd, 2011; for discussion on complex metazoan multicellularity, see Nejad Kourki, 2022).14

Although organismal complexification is one way in which evolution proceeds to produce variation in particular lineages (Bonner, 1998; McShea, 2016), it is not the only way. The role of simplification is becoming increasingly recognised as a major driver of trait diversity within Evo-Devo and evolutionary biology more generally (e.g., O’Malley et al., 2016; Schoch, 2014). Secondary simplification, as the term suggests, refers to the idea that the evolution of traits as observed in some extant species may have involved a trajectory from complex to simple or from more complex to less complex traits. Such simplification may involve genetic streamlining (i.e., deletion of genes or gene complexes) and/or the gradual loss of anatomical structure or behavioural phenotypes (O’Malley et al., 2016). One example of interest for Evo-Devoists of how secondary simplification has played a role in evolutionary diversification can be found when looking at the phylogenetic relationship between choanoflagellates, filastereans, and metazoans.

Choanoflagellates are currently recognised as the closest living relative of metazoans and this makes them of particular interest to biologists investigating the transition from unicellularity to multicellularity (King et al., 2008; Ruiz-Trillo et al., 2023), as we mentioned in Sect. 3. One of the various routes to multicellularity that has been hypothesised (there are at least 11 for each of the major eukaryotic clades), suggests that animals originated from a colonial stage reminiscent of present-day choanoflagellates (for other alternatives, see Ruiz-Trillo et al., 2023).15 Genomic comparison between choanoflagellate Monosiga brevicollis and various animals has been used to infer that their last common ancestor shared some cell adhesion and cell–cell communication gene complexes but failed to possess many other significant signalling pathways and transcription factors found in animals and important to their multicellular development (O’Malley et al., 2016). However, recent genomic sequencing of Capsaspora owczarzaki, a unicellular opisthokont belonging to Filasterea—a close sister clade to both choanoflagellates and animals, has revealed that they share developmentally important cell adhesion genes complexes (i.e., integrins) and transcription factors with animals which were lost in choanoflagellates (Sebé-Pedrós et al., 2010). This has led biologists to suggest that a secondary simplification may have occurred. It is likely that the last common ancestor of filastereans, metazoans, and choanoflagellates possessed integrins and various transcription factors that were lost in choanoflagellates but retained and repurposed in metazoans.

Many more examples of Evo-Devo research and discussions on secondary simplifications could be cited (e.g., Hannibal & Patel, 2013; Hoekstra et al., 2018; Virágh et al., 2021), including cases related to nervous system evolution in metazoans (Roth et al., 1997; Tosches, 2017). The implications of secondary simplifications are an underexplored topic in basal cognition research, a programme which, as we saw in the previous section, has been grounded on an assumption of gradualism and complexification over evolutionary timescales; there is a tendency to think in linear terms.

One additional moral of the story, to paraphrase Clark (1999), is that ‘simple’ does not always mean evolutionarily primitive (see also Godfrey-Smith, 2017). In fact, as some Evo-Devoists have argued, basal taxa are not necessarily the best available stand-ins for ‘ancestors’ (Jenner, 2006), and these, due to being adjacent to the root or in proximity to it in a (rooted) phylogenetic tree, are often incorrectly interpreted as having undergone less change and thought to be more “primitive” or “ancestral” than subsequently branching lineages (Crisp & Cook, 2005; Gregory, 2008). It may be beneficial for basal cognition researchers to ponder this when selecting their model and study organisms.

Flipping the coin and chewing over complexification from the viewpoint of major transitions in individuality that we sketched above, as these also need to be weighed in, what might the processes of vertical complexification mean for basal cognition? In any given instance, should we shift the locus of analysis to the higher biological level under the assumption that it becomes the more ‘dominant’ level of selection that constrains the behaviour of the lower levels, or should we continue to focus on lower levels despite their being less autonomous? In other words, major transitions in individuality are conceptually and empirically distinct from major transitions in cognition and although the later have been explored in detail (Barron et al., 2023; Ginsburg & Jablonka, 2019, 2021), the implications of the former have gone under the radar to the detriment of basal cognition.

Complexification and simplification are just two evolutionary scenarios that involve gains and losses of traits in evolutionary history. However, the parallel that we are delineating for basal cognition research reaches beyond these scenarios. We want to underscore the overall importance of gains and losses of traits in evolutionary investigations, something that deeply affects the types of inferences and explanations that could (or shouldn’t) be drawn from the species chosen for scientific study. In this domain, Evo-Devo has many lessons to bestow to basal cognition research. For one, Evo-Devoists argue that both losses and gains are important for explaining evolution (e.g., Cannell et al., 2020).16 Evolution is thus sometimes described, from a phylogenetic perspective, as a “never-ending cycle of gains and losses” (Keller & Delaux, 2022). In this tenor and building from the corpus of knowledge on the evolution of other biological features, we forecast that both gains and losses will prove to be salient when addressing the evolution of certain cognitive traits present in extant organisms (and perhaps for the evolution of some of the cognitive capacities in the more encompassing basal toolkit).

Furthermore, there are a few extra things to bear in mind. Evo-Devoists have argued that we should fight our unintentional bias toward interpretations that favour trait gains over losses (Church & Extavour, 2020) and have shown than the loss of traits is not an all-or-nothing phenomenon, but rather traits are lost to varying degrees along a so-called “trait loss continuum” (Sadier et al., 2022). Likewise, multiple gains and losses of characters in related taxa are not at all uncommon, something that could be partly accounted for through the Evo-Devo notion of ‘deep homology,’ “pointing to the persistence of developmental potentials that are not always expressed” (DiFrisco, 2019). The loss and gain of cognitive characters could also be subject to evolvable potentials that are not always expressed, but that could easily tilt in one or the other direction in related taxa. In sum, the loss and gain of traits (including complexification and simplifications) are intricate explananda and simultaneously powerful explanantia in fully evolutionary fields. Recognising their significance and giving them adequate treatment would help ensure that basal cognition research develops in a manner that is consistent with being a fully evolutionary approach.

The final remark that we would like to advance on this theme is that countenancing losses and gains in cognitive evolution forces us to reckon with the fact that in some lineages, and only in those lineages, there might be cases of unique traits (i.e., autapomorphies). In this sense, a fully evolutionary research program should not avoid questions about cognitive uniqueness, for instance, in Homo sapiens, but also in other non-human organisms, including non-neuronal ones. As an illustration, if enough evolutionary evidence allows us to call (certain) fungi ‘cognitive,’ they might be unique in different ways that, for instance, birds or humans might fail to be. Regarding this issue, we think that basal cognition researchers could follow Shettleworth (2010), who claims that in the evolutionary study of cognition

[…] neither blanket anthropomorphism nor complete anthropodenial is the answer […]. When it comes to comparing human cognition with that of other species, it is most likely that—just as with our genes and other physical characters—we will find some processes shared with many other species, some with only a few, and some that are uniquely human. (p. 19)

As we mentioned in Sect. 2, the (Evo-Devo-inspired) Yin and Yang of investigating conservation and difference in evolution, including gains and losses of traits that make certain lineages stand out, should be central for basal cognition research.