The Phanerozoic aftermath of the Cambrian information revolution: sensory and cognitive complexity in marine faunas

Abstract

The Cambrian information revolution describes how biotically driven increases in signals, sensory abilities, behavioral interactions, and landscape spatial complexity drove a rapid increase in animal cognition concurrent with the Cambrian radiation. Here, we compare cognitive complexity in Cambrian and post-Cambrian marine ecosystems, documenting changes in animal cognition after the initial Cambrian increase. In a comparison of Cambrian and post-Cambrian Lagerstätten, we find no strong trend in the proportion of genera possessing two types of macroscopic sense organs (eyes and chemoreceptive organs such as antennae, feelers, or nostrils). There is also no trend in general nervous system complexity. These results suggest that sophisticated information processing was already common in early Phanerozoic ecosystems, comparable with behavioral evidence from the trace fossil record. Most taxa capable of complex information processing in Cambrian ecosystems were panarthropods, whereas mollusks and chordates made up larger proportions afterward. In both the Cambrian and the present day, ecological occupation of diverse habitat tiers and feeding modes is possible with even simple nervous systems, but ecological lifestyles requiring rapid, regular movement are almost exclusively associated within brain-bearing taxa, suggesting a connection with fast information-processing abilities and bodily responses. The overall rise in cognitive sophistication in the Cambrian was likely a unique event in the history of life, although some lineages subsequently developed more elaborate sensory systems and/or larger brains.

Introduction

Cognition is the ability to acquire, process, and respond to information. The way animals interact with physical and biotic environments is heavily mediated through use of information, such that innovations and changes in their cognitive abilities can greatly influence their overall ecology and evolution (Dukas and Ratcliffe 2009). The origins and evolution of cognition are of intense interest to biologists, but they have been rarely examined in the fossil record.

The Cambrian radiation of metazoan animals is a critical event for understanding the origins of cognition. In the Cambrian, the world became much more complicated to navigate as the overall size, diversity, disparity, and variety of lifestyles of macroscopic animals increased (Valentine 2002; Marshall 2006; Dunne et al. 2008; Bush et al. 2011). The spatial landscape was also transformed during the Ediacaran and Cambrian; large concentrations of biomass and organic matter produced heterogeneity (the “savannah hypothesis”; Budd and Jensen 2015) and seafloor sediments were increasingly modified by bioturbators, disrupting the stable matground surfaces that existed before (the agronomic and substrate revolutions; Seilacher and Pflüger 1994; Bottjer et al. 2000). There were benefits to being able to collect, process, and respond to information in this new, spatially and biotically complex world, such as acquiring difficult to locate resources. There were also costs borne from missing out on information, including risks of undetected danger. Therefore, selection pressures existed for increased abilities to handle and utilize information—in other words, cognition—among mobile Cambrian bilaterian metazoans. This idea was proposed by Plotnick et al. (2010) as the Cambrian information revolution.

Though the cognition of animals is not directly preserved in their fossils, morphology and other correlates provide clues about how animals handled information. We can break down cognition into three parts—collecting, processing, and responding to information. First, an animal's ability to collect information from its surroundings can be determined from its sensory systems, reflected in the presence, or degree of development, of sense organs. Second, an animal's ability to process information can be constrained from its nervous system, remains of which have been found, albeit rarely, in well-preserved fossils. Third, how an animal potentially responded to information is revealed by its behavior, which is inferable from functional morphologic analysis and from trace fossils, the preserved results of behavior. These aspects may also be inferred phylogenetically from modern living relatives of the animal. Together, these lines of evidence allow us to reconstruct cognition and cognitive abilities in deep time.

From the sensory side, Plotnick et al. (2010) and Zhao et al. (2013) examined the Cambrian information revolution by tabulating the number and proportion of animals with macroscopic sense organs (eyes and antennae) in the Chengjiang biota, finding that they disproportionately belonged to active, mobile animals. Hunting or scavenging animals, as well as epifaunal, pelagic, and especially nektonic organisms, were most likely to have eyes. Examples of this trend are seen with the large and well-developed eyes of the swimming arthropods Isoxys, Tuzoia, and Waptia (García-Bellido et al. 2009; Strausfeld 2011, 2016; Vannier et al. 2018). Parker (1998, 2003) proposed that the development of eyes and vision triggered the Cambrian radiation, generating a diversity of body plans from arms races between competing organisms that could see one another, though Plotnick et al. (2010) consider that complex eyes evolved too late for this. Nonetheless, the senses involved in the information revolution, including vision, chemoreception, and mechanoreception, likely promoted increased biotic interactions, evolution, and diversification in the Cambrian and later times.

Evidence of sophisticated information collection and processing comes from multiple studies of exquisite preservation in Cambrian fossils of complex eyes, other sensory organs, nervous systems, and brains (e.g., Chen et al. 1999; Shu et al. 2003; Paterson et al. 2011; Ma et al. 2012; Schoenemann and Clarkson 2013, 2017; Tanaka et al. 2013; Cong et al. 2014; Strausfeld 2015). Panarthropods (arthropods and their relatives), possessing a diversity of eyes and antennae and with brains resembling those of their modern counterparts (Ma et al. 2012; Tanaka et al. 2013; Cong et al. 2014; Strausfeld 2015), were particularly dominant among the active and visually oriented fauna during this period (Zhao et al. 2013). Other phyla known to be neurologically and sensorially complex today were also well represented at this time. Cambrian annelids bearing organs such as antennae and paired sensory palps included Kootenayscolex (Nanglu and Caron 2018) as well as Canadia, which was also found with an exceptionally well-preserved nervous system, suggesting that the ancestor of annelids was already reasonably cephalized (Parry and Caron 2019). The enigmatic nectocaridids, interpreted as nektonic cephalopod-like mollusks, possessed large, camera-type eyes and flexible tentacles (Smith 2013). Sensory evidence in Cambrian chordates ranges from paired tentacles in the weakly cephalized Pikaia (Morris and Caron 2012) to more fish-like traits in animals interpreted as craniate or vertebrate, such as the eyes, possible nasal sacs, and potential otic capsules seen in Haikouichthys (Shu et al. 2003). Likewise, the chordate Haikouella was found with a sizable preserved brain that appeared to be tripartite (Chen et al. 1999). Cambrian early vertebrates and their relatives may have already possessed brain divisions seen in living vertebrates, such as the hindbrain, midbrain, and forebrain with cerebral hemispheres (Shimeld and Holland 2000; Northcutt 2002; Chen 2008).

The Cambrian provides us with the first definitive evidence of brains in the fossil record, although neurons and nervous systems would have been present earlier. Paulin and Cahill-Lane (2021) suggested that neurons evolved from epithelial cells in Ediacaran mat grazers, which first developed a need to process and respond to event-based sense data resulting from innovation in feeding and motility. Decentralized nervous systems in the form of diffuse nerve nets probably characterized much of the Ediacaran fauna according to Northcutt (2012), with ganglionated nervous systems possibly approaching brain-level organization for some animals with apparent cephalization (e.g., Spriggina), especially if they turn out to represent clades related to annelids or panarthropods. Northcutt (2012) also discusses the unresolved question of whether the last common bilaterian ancestor possessed a centralized brain or whether it had a simpler nervous system from which brains evolved multiple times independently. Likewise, Evans et al. (2021) consider, based on body plan–related characters in Ediacaran macrofossils, that the rudimentary nervous systems in early animal lineages persisted for a long while before separate developments of a central nervous system.

Feinberg and Mallatt (2013) and Barron and Klein (2016) argued that consciousness first appeared in the Cambrian when, with the onset of advanced brains, animals could neurally represent the external world and perceive themselves moving within it. Similarly, according to Trestman (2013), the diversification of animals with complex, active bodies during the Cambrian radiation itself was dependent on “embodied cognition”—spatial awareness and bodily perception and control—which allowed for agent-like actions, such as manipulating objects.

Behavioral evidence provides another view of the Cambrian information revolution, revealing how animals responded to received information from the environment and other individuals, and how they may have sent information to others in return. Numerous types of behaviors have been attributed to Cambrian animals (Table 1). Some of these lines of evidence reflect inferences from morphological adaptations to life modes (e.g., predation) that require certain behaviors, or features such as color patterns or ornamentation meant to signal and influence the behaviors of others. Other lines of evidence reflect the results of behavior as preserved in trace fossils or in body fossils in life positions (e.g., hiding in enclosed spaces).

Table 1. Behaviors, behavioral abilities, or morphological correlates of behavior inferred to exist by the Cambrian.

Though animals were already fully motile, leaving locomotion traces, in the Ediacaran (Evans et al. 2019; Ivantsov et al. 2019); the trace fossil record also speaks to a revolution in sensing, cognition, and behavior across the Ediacaran/Cambrian boundary. Carbone and Narbonne (2014) examined strata from northwest Canada spanning this interval, finding that trace fossils representing only simple sensory behaviors, like undirected horizontal traces and two-dimensional avoidance traces, dominated the Ediacaran, while the succeeding parts of the Cambrian (Terreneuvian) yielded traces showing more sophisticated, complex feeding behaviors like zigzag probing and vertical probing. A well-known example is the burrow Treptichnus, which defines the boundary between the Ediacaran and Cambrian periods (Vannier et al. 2010; Buatois 2018). Based on fossil to modern comparisons, treptichnids like Treptichnus have been linked to priapulid tracemakers (Vannier et al. 2010; Kesidis et al. 2019) that have nerve networks arranged in rings and longitudinally (Adrianov and Malakhov 2001); such nervous systems existing in the late Ediacaran are consistent with the ideas of Northcutt (2012) and Evans et al. (2021).

Following up on these studies, we explore how Cambrian faunas compare with later marine faunas in sensing and cognition to test the hypothesis that the Cambrian information revolution was a unique event in the history of metazoans. It may be that Cambrian ecosystems already achieved comparatively modern levels of complexity, with later changes being more modest. Alternatively, the sensory and cognitive complexity of animals in marine ecosystems may have continued to rise considerably, for a number of reasons. Heterogeneity that favors cognition (Plotnick et al. 2010) may have increased further as ecosystem structure became yet more complicated, with greater tiering (Bottjer and Ausich 1986; Droser and Bottjer 1989; Bush et al. 2007) and with the proliferation of complex, three-dimensional habitats such as reefs (Wood 1999). Through the Phanerozoic, biodiversity increased, which might lead to increased biotic interactions through higher rates of different kinds of animal detecting, encountering, and responding to one another, and thus an increased demand for cognition. For example, the great Ordovician biodiversification event produced enormous taxonomic and ecological diversification within phyla (Servais and Harper 2018), and Mesozoic radiations produced a “modern” fauna more diverse and ecologically complex than what existed before (Sepkoski 1981, 1984; Bush and Bambach 2015), including the sensorially and behaviorally rich vertebrates. All the while, the average biomass and metabolism of animals increased, energetic life modes became prominent (including mobile, specialized predators, and infaunal burrowers), and new kinds of escalation between predators and prey took place (Vermeij 1977, 1993; Bambach 1983, 1993; Bush et al. 2007; Knope et al. 2015; Smith et al. 2016), which could favor better sensory processing among metazoans alongside better physical offensive and defensive strategies.

We examine whether and to what extent the share of animals with complex sensory or information-processing systems increased in post-Cambrian marine ecosystems after the initial information revolution wherein many of these systems first appeared. We thus will place the Cambrian information or sensory revolution in the context of the rest of the Phanerozoic. We examine this question from three points of view. First, we compare the proportion of animals with macroscopic sense organs reflecting photoreception (eyes) and chemoreception (antennae, feelers, or nostrils) in well-known marine Lagerstätten from the Cambrian and post-Cambrian, as well as one modern marine fauna (cf. Plotnick et al. 2010; Zhao et al. 2013). Possession of both types of (typically paired) organs allows for increased sensory acuity as well as spatial perception of the stimulus (Plotnick et al. 2010). Lagerstätten were chosen because their high-quality preservation allows a wider taxonomic census; they are also heavily researched and have readily available published genera or species lists in academic papers and/or field guidebooks.

Second, for the same set of faunal assemblages, we infer the proportion of animals with four different levels of nervous system complexity: (1) no neurons, thus no nervous system; (2) decentralized nervous system (e.g., nerve net or ring); (3) nervous system with ganglia, but no brain; and (4) nervous system with a centralized brain. These levels represent grades in organization of information processing and can be applied to fossil taxa by comparison with modern analogues and their body plans.

Finally, we determine which ecological life modes and areas of ecospace are associated with these four grades of neurological complexity using the scheme of Bambach et al. (2007) and Bush et al. (2007). This allows us to consider whether certain lifestyles require or favor more well-developed information-processing systems than others, both in the Cambrian and today.

Methods

Lagerstätten Faunas

We compared 23 faunal lists, 22 from famous and well-described marine Lagerstätten (14 Cambrian, 8 post-Cambrian) and one from a guidebook representing a modern marine fauna (Bermuda). Fossil faunal lists came from the published literature and field guidebooks (see Supplementary Data). For each faunal list, we tallied the proportion of metazoan genera present that had or are inferred to have had macroscopic photoreceptive organs (eyes) and macroscopic chemoreceptive organs (antennae, feelers, or nostrils).

The 14 Cambrian Lagerstätten faunal lists include the major Burgess Shale–type (BST) biotas—the Chengjiang, Sirius Passet, Sinsk, Guanshan, Balang, Emu Bay, Kinzers, Kaili, Spence, Burgess Shale, Wheeler (House Range), Wheeler (Drum Mountains), Marjum and Weeks (Foster and Gaines 2016; Holmes et al. 2018). The post-Cambrian Lagerstätten include the Ordovician-aged Fezouata (Van Roy et al. 2015) and Soom Shale (Gabbott et al. 2017), Devonian Hunsrück (Südkamp 2017) and Carboniferous Mazon Creek (Wittry 2012), the Jurassic La Voulte-sur-Rhône (Charbonnier et al. 2014), Oxford Clay (Martill and Hudson 1991) and Solnhofen (Bartel et al. 1990), and the Eocene London Clay (Rayner 2009). The one modern faunal list comes from Bermuda (Sterrer and Schoepfer-Sterrer 1986), a region well studied and surveyed across all major taxonomic groups by generations of marine biologists (Beers 1963). This well-described contemporary ecosystem provides a useful comparison to our well-described fossil ones. Although these assemblages differ in preservational mode and history of study, which adds inevitable variability to our analyses, we consider using these well-studied Lagerstätten to be the closest to lifelike censuses of ecosystems possible among fossil assemblages, including many taxa that would otherwise not be preserved.

We restricted our analysis to marine organisms, except for the Mazon Creek, for which the freshwater to saltwater transition is not clearly defined. For this Lagerstätte, we included all aquatic taxa. Overall, we included marine reptiles, marine mammals, and amphibians, but excluded flying seabirds or pterosaurs. For consistency with the fossil assemblages, we also excluded the small phyla Tardigrada, Gnathostomulida, Gastrotricha, Rotifera, and Kinorhyncha, as well as internal parasites (e.g., tapeworms, acanthocephalans), from the modern Bermuda fauna. These animals had limited preservational potential in our fossil assemblages, although there may be occasional examples of them elsewhere in the fossil record. Trace fossil taxa were also excluded. The number of genera used in each Lagerstätte for each of the anatomical comparisons is shown in Table 2.

Table 2. Total number of genera in each Lagerstätte or other faunal assemblage examined and number of genera that could be scored for the absence or presence of eyes, chemoreceptive organs, and type of nervous system. Some taxa were excluded from scoring based on reasons such as unclear preservation or uncertain or disputed taxonomic assignment.

Sensory Organs

Animals were coded as having eyes if there were macroscopic visual organs. Macroscopic chemoreceptive organs include the antennae of arthropods and polychaetes, the paired feelers or tentacles of gastropods, and the nostrils of chordates. Because many of these anatomical traits are rarely preserved, several lines of evidence were used for the coding, both direct and indirect. Anatomical traits were examined from descriptions in published papers and guidebooks, including any photos, figures, or reconstructions. If these were not available, traits were assigned to the fossil genera within crown groups known to possess them in the modern (e.g., crustaceans have antennae, except for known cases where they have been lost, like barnacles; all sponges lack sense organs or nervous systems). Taxonomic information for such coding was generally taken from the published sources of the faunal lists if available, or from online sources such as the Paleobiology Database, though this was not done for highly disputed or uncertain taxonomic assignments. Genera for which the anatomical assignment was too uncertain were excluded from the analyses (e.g., unclear preservation of the feature or belonging to extinct phyla without clear modern analogues). These make up a noticeable minority of nearly all analyzed Cambrian and Ordovician faunal lists (e.g., typically 10%–20%, but up to slightly over a third), but had limited effect on the later faunas (uncertain coding only existed in Hunsrück and Mazon Creek genera, making up <10%, and mostly in annelid sense organs).

Nervous System Complexity

We also classified metazoan taxa in the same set of faunal assemblages into four major categories of nervous system complexity based on the published literature and invertebrate zoology textbooks. The first category includes multicellular organisms without nervous systems (i.e., sponges). Although they are able to respond to stimuli, they do not have specialized information-relaying cells (i.e., neurons) that directionally pass electrochemically encoded signals around the body much faster than typical chemical diffusion (Nickel 2010). The second category includes animals with decentralized nervous systems, like cnidarians and echinoderms. Neurons are positioned in tracks reaching to and from various parts of the body (nerve nets or rings). The third category includes animals that possess ganglia, which are clusters or bundles of nerve cells that help organize sensing and action by controlling different processes or different areas of the body. The fourth category includes animals in which the nervous system is further organized and centralized into a major organ, the brain, which is located anteriorly and often near major sense organs. The brain sends and receives signals to and from the rest of the body, which is thought to be necessary for coordinating complex, active bodies with large behavioral repertoires (Trestman 2013).

As with sense organ anatomy, but even more so, nervous systems are only very rarely directly fossilized, so we have relied on inference based on modern analogues and representatives of taxonomic groups. Body plans are highly conserved among phyla or classes within phyla, and nervous system organization, at least on a broad level, is often tightly correlated with aspects of body plans (Evans et al. 2021)—for instance, possession of an anterior-posterior axis, cephalization, appendages, and muscles. Most of our fossil genera could be assigned to at least the phylum level, where these traits are inferable, though the share is smaller for Cambrian ones, as discussed later on. Nonetheless, we acknowledge these types of paleobiological analyses will always have some uncertainties, although the same is true of many other studies. Because nervous system complexity is tightly correlated with phylogeny, trends in nervous system complexity will reflect trends in the taxonomic composition of the fauna. However, this does not mean that these trends are not worthy of examination. Nervous system complexity influences an animal's ability to manipulate its environment, interact with other species, and respond to stimuli, so these trends have important implications for ecology and evolution, even if they are tied to taxonomy. A similar approach has been used in studies of respiratory and circulatory system organization (Heim et al. 2020), vision (Aberhan et al. 2012), reproductive mode (Bush et al. 2016), and even life modes such as motility (Bush et al. 2007).

Brains were inferred in the chordates, arthropods, stem-arthropods, lobopodians, cephalopod and gastropod mollusks, and some annelids. Among annelids, polychaetes having sense organs and active life modes were generally considered to have brains, whereas other annelids generally were coded as ganglia-only; this is a judgment based on the complexity of the nervous system in members of the phylum today (Beesley et al. 2000; Rouse and Pleijel 2001; Orrhage and Müller 2005; Heuer et al. 2010). Additionally, to examine the taxonomic distribution of brains in each fauna, we calculated the relative proportion of brain-bearing genera belonging to the following taxa: chordates, panarthropods (arthropods and their relatives, including lobopodians), cephalopods, gastropods, and annelids.

Statistical Comparison

To test whether Cambrian and post-Cambrian assemblages were different in proportions of eyes, chemoreceptive organs, and brains, we performed a two-step resampling routine run with 10,000 iterations. We resampled the genera in each assemblage with replacement at the original sample size (number of genera), calculating the proportion possessing the anatomical trait for each iteration. Then, to calculate the mean proportion for each interval (Cambrian and post-Cambrian), we resampled the assemblage-level means with replacement at the original sample size (number of assemblages). The Cambrian mean was subtracted from the post-Cambrian mean for comparison—there would be a significant difference between the two intervals if the 95% confidence interval did not overlap zero.

Life Modes and Nervous System Complexity

To examine the relationships between our four levels of nervous system complexity (no neurons, decentralized system, ganglia, or brain) and ecological modes of life, we used the ecospace scheme of Bambach et al. (2007), which categorizes marine animals with respect to three parameters—tiering relative to the sediment–water interface, motility level, and feeding mechanism. Six possible values for each parameter exist, with each combination of the three parameters defining a unique life mode; the resulting ecospace is depicted as a cube (Fig. 1). Not all of the 216 theoretical life modes are employed by known organisms, such that not all of the theoretical ecospace is actually filled. Bambach et al. (2007) and Bush et al. (2011) examined which parts of ecospace were actualized in the Recent and at other times in the history of animal life.

Figure 1. Theoretical ecospace of marine animals, from Bush et al. (2007), with the three ecological parameters of tiering, motility level, and feeding mechanism. Definitions of the individual categories within each parameter are also in Bambach et al. (2007) and Bush et al. (2007).

We use the data in Bambach et al. (2007) and Bush et al. (2011) to evaluate which life modes (and thus extent of ecospace occupation) are associated with our four types of nervous system in animals in the Recent and the Cambrian. The data were restricted to the major phyla, including the larger soft-bodied “worm” phyla but excluding minor, small-bodied taxa such as rotifers and placozoans. The taxa we included for each level are as follows: (1) no nervous system—Porifera; (2) decentralized nervous system—cnidarians, ctenophores, echinoderms, hemichordates, priapulids, and brachiopods; (3) ganglia—bryozoa, non-gastropod, non-cephalopod mollusks, non-polychaete annelids (including pogonophorans and echiurans in the annelids), sipunculids, nemerteans, “platyhelminth” flatworms, and nematodes; and (4) brain—all chordates, panarthropods, gastropods, cephalopods, and polychaetes. This is generally similar to how they were coded for our Lagerstätten comparison analysis. Cambrian taxa excluded because they could not be assigned to a nervous system type include anabaratids, cambroclavids, chancellorids, coeloscleritophoans, decollating tubular fossils, halkierids, hyolithelminthes, hyoliths, protoconodonts, stenothecoids, tommotids, vetulicolians, and trace fossils, plus those labelled “problematica” and not assigned to a phylum. These excluded taxa had 15 life modes among them, but of these, all but two (possessed by anabaratids and decollating tubular fossils) were inhabited by Cambrian taxa included in the analysis.

Results

Lagerstätte Faunas

The proportion of genera bearing eyes (Fig. 2A) across the 19 faunas ranged from 24% (Sirius Passet) to 83% (London Clay). On average, across all faunas, around half of the genera had eyes. Cambrian faunas had 45% of genera with eyes on average, compared with 62% in the post-Cambrian. The average difference between the post-Cambrian and Cambrian was 18%, which was statistically significantly different from zero, with a confidence interval of [3%, 31%]. Thus, there was a significant increase in the mean proportion of eyes.

Figure 2. Proportion of listed genera in each fauna possessing macroscopic sense organs. A, Eyes. B, Chemoreceptive organs—antennae, feelers, or nostrils.

The proportion of genera bearing macroscopic chemoreceptive organs (antennae, feelers, or nostrils) across the 19 faunas ranged from 22% (Hunsrück) to 82% (Weeks) (Fig. 2B). On average, across all faunas, 48% of the genera had these chemoreceptive organs. Cambrian faunas had 45% of genera with these organs on average, compared with 51% in the post-Cambrian. The mean percentage of genera with chemoreceptive organs was not statistically significantly different in the post-Cambrian and Cambrian, with the confidence interval on the difference being [−10%, 20%].

In most faunas, the majority of genera had centralized brains—on average 65% in the Cambrian faunas and 66% in the post-Cambrian (Fig. 3A). The difference between the post-Cambrian and Cambrian was not significant, with a confidence interval of [−12%, 14%]. The average share of genera with ganglion-level organization was lower in the Cambrian (<1%) than the post-Cambrian (11%), with bivalves and bryozoans contributing heavily to the post-Cambrian value. The low Cambrian value might partly reflect uncertainty in assigning genera to this category compared with the three others. Decentralized nervous systems were possessed by 22% of the Cambrian faunas on average, with a similar 21% in the post-Cambrian. Finally, 12% of genera had no nervous system in the average Cambrian fauna, reflecting a diversity of sponges, compared with only about 2% in the post-Cambrian.

Figure 3. Nervous system complexity. A, Proportion of listed genera in each fauna by level of nervous system complexity. B, Proportion of genera that possess brains that belong to each major taxonomic group.

Even though the average proportion of genera with brains was similar in Cambrian and post-Cambrian faunas, there is a notable difference in which taxonomic groups make up the genera with brains (Fig. 3B). Cambrian genera with brains are overwhelming panarthropods (89%–100%), with the remaining genera being annelids or chordates. Nine of the 14 Cambrian faunas have panarthropods as the only animals with brains. By contrast, the post-Cambrian faunas surveyed are more diverse in brain-bearing taxa—the majority have representatives of the chordates, panarthropods, cephalopods, gastropods, and annelids. The transition to this more diverse set of brain-bearing taxa can be seen in the two Ordovician faunas. In the Early Ordovician Fezouata biota, 83% of genera with brains are still panarthropods, but small shares of chordates, cephalopods, gastropods, and annelids are also present. Later in the Ordovician, only 57% of brain-bearing genera are panarthropods in the Soom Shale. While a single major taxon dominates the brain-bearing genera in some post-Ordovician faunas (i.e., La Voulte-sur-Rhône and Hunsrück have slight panarthropod majorities, and the London Clay and Solnhofen have slight chordate majorities), no group overwhelmingly dominates among the brain-bearing taxa in these later faunas the way panarthropods did in the Cambrian and Early Ordovician.

Life Modes and Nervous System Complexity

Among the major taxa we examined, 6 ecological modes of life are occupied in the Recent by genera with no nervous system, 47 by those with decentralized nervous systems, 41 by those with ganglia, and 60 by those with a brain (Fig. 4). Thus, metazoans with no nervous system (i.e., Porifera; Fig. 4A) occupy much less ecospace than metazoans with a nervous system of any kind (Fig. 4B–D). The few life modes occupied by the former group are characterized by a lack of motility, and they occupy only a few tiers and a few feeding modes, mainly suspension feeding. With even a decentralized nervous system (Fig. 4B), animals can move around, occupying all but the most active of the six motility categories (fully motile fast); they can also occupy all six tiers and all six feeding categories. Animals with ganglia are similar in this range of ecospace occupation (Fig. 4C). Much of this ecospace continues to be used by animals with brains (Fig. 4D), but only the brain-bearing animals occupy the “fully motile fast” category (Fig. 4D, front rows closest to viewer in diagram). This category involves regular rapid and unencumbered movement, such as walking or swimming (and which differs from “fully motile slow,” whereby routine movement retains more of a bond with substrates, such as creeping or gliding). A borderline exception may be the chaetognaths (not included in figures), small fast predators on plankton that have existed since the early Cambrian (Vannier et al. 2007), which we might code as having ganglia rather than a well-developed brain, because their cerebral ganglion is one of multiple ganglia in the central nervous system (Harzsch and Wanninger 2010).

Figure 4. Life modes associated with each of the four levels of nervous system complexity in the Recent. A, No nervous system. B, Decentralized nervous system. C, Ganglia. D, Brain.

In the Cambrian (Fig. 5), 4 ecological modes of life are occupied by animals with no nervous system), 17 by those with a decentralized nervous system, 8 by those with ganglia, and 15 by those with a brain. Although there are fewer modes of life overall (as seen in Bambach et al. 2007; Bush et al. 2011; Knope et al. 2015), the relationship between nervous system development and ecospace occupation that is found in the Recent is already developed. Specifically, metazoans without neurons are associated only with the lowest motility category and with limited tiering and feeding mechanisms; possessing a nervous system—even a decentralized one—is linked to major expansion into additional motility levels, tiers, and feeding modes. Also, brains are associated with the occupation of the fully motile fast life modes.

Figure 5. Life modes associated with each of the four levels of nervous system complexity in the Cambrian. A, No nervous system. B, Decentralized nervous system. C, Ganglia. D, Brain.

There was an ecospace increase between the Cambrian and Recent for all four levels of neurological complexity, reflecting the continued Phanerozoic ecological diversification of many metazoan phyla past their initial Cambrian appearances. Though only slight increase is seen in the neuron-less sponges, the three categories possessing nervous systems all more than double their number of life modes. The close to tripling of life modes for decentralized nervous systems is associated with the ecospace expansion of cnidarians and echinoderms, and the fivefold and fourfold rise in life modes for ganglia and brains, respectively, is associated with those of numerous other bilaterians. Though there is some difference and lag in which regions of ecospace were occupied first (for instance, animals with decentralized nervous system reach all tiers in the Cambrian, earlier than those with ganglia or brains do), by the Recent, all three nervous system categories end up occupying very similar regions of ecospace aside from the additional region of “motile fast” ecospace held by brains alone.

Arthropods were the most varied in life mode among the groups with brains, although all brain-bearing taxa ecologically diversified to a major degree, with many of their life modes overlapping. In the Cambrian, arthropods occupied 10 of the 15 life modes possessed by animals with brains, polychaetes occupied 3, while gastropods, cephalopods, and chordates each occupied only 2 life modes. Among the 60 Recent life modes occupied by animals with brains, arthropods possess 42; chordates, 26; polychaetes, 23; gastropods, 15; and cephalopods, 4.

From the Cambrian to the Recent, the number of life modes in the most active, fast-moving category, which as previously mentioned are nearly exclusively linked to brains, nearly quadrupled, from 4 to 15. Arthropods held all (and chordates held 1) of the 4 Cambrian fast life modes, while more groups comprised the 15 Recent fast life modes—arthropods (13 modes), chordates (9 modes), cephalopods (2 modes), and polychaetes (1 mode).

Discussion

Our results support the idea that Cambrian ecosystems were very “modern” in the proportion of animals capable of information processing as reflected through their anatomy, comparable to those later in the Phanerozoic. Sense organs were abundant across most of the faunas we examined, with post-Cambrian ecosystems on the whole averaging more, particularly for eyes, due to the addition of cephalopods and chordates as major components of the fauna. Cambrian and post-Cambrian faunas had similar percentages of genera with brains, with all but one fauna having more than 50%, implying marine ecosystems had high cognitive complexity through much of the Phanerozoic. The high proportion of both types of sense organs alongside brains is expected, as these traits covary—brains are needed to process input from the senses, and vision and chemoreception can be strongly complementary for many animals engaging in active navigation (Plotnick et al. 2010). In the Cambrian, as in the present, the presence of a nervous system in marine animals was associated with occupation of diverse habitat tiers, feeding modes, and motility levels. However, the most active life modes—those classed as fully motile fast—are almost exclusively associated with brain-bearing animals, suggesting such lifestyles require the fast information-processing abilities and muscular responses associated with a centralized nervous system. Information-processing ability is expected to be a major constraint on the potential to enter or radiate within certain regions of ecospace, even if many life modes can be occupied by animals with varying levels of neurological complexity. For instance, epifaunal grazer roles can be filled by eyeless, brainless sea urchins as well as cephalized snails, while nerveless sponges and ganglionated bivalves alike act as reef builders in different time periods. However, no fast-moving swimming predators in the manner of anomalocarids or dolphins exist among the brainless. In the Cambrian, complex sensory and nervous systems were concentrated in a single major taxon, the (pan)arthropods, whereas later faunas had a more diverse set of sensorially and cognitively complex groups, with chordates and mollusks becoming important constituents of the fauna.

Cambrian and post-Cambrian faunas overlap heavily in the proportion of genera with sense organs or brains. Some of the variation in these proportions may be habitat related; for example, the two faunas with the lowest share of animals with eyes represent deeper-water habitats. The Sirius Passet fauna is reconstructed as having lived below the photic zone (Hammarlund et al. 2019; Harper et al. 2019), and the Hunsrück Slate was deposited at the least below storm wave base (Brett and Seilacher 1991; Sutcliffe et al. 1999). The Hunsrück was also the only one we analyzed in which a majority of the genera were inferred to lack brains.

Our comparison of Cambrian and post-Cambrian faunas is complicated by the exclusion of genera with uncertain anatomical coding, which are concentrated in the Cambrian. Our results may be biased in favor of greater Cambrian sensory and nervous system complexity if the excluded uncertain genera disproportionately lacked the more complex traits. If we reanalyze the data with all uncertain codings set to represent lack of eyes, antennae, and brains (which we consider very unlikely), the average faunal list in the Cambrian drops from 45% to 40% of genera having eyes, from 45% to 40% having the inferred macroscopic chemoreceptive organs, and from 65% to 54% having brains. This nonetheless still represents a large share with complex sensory and nervous systems, showing the importance the Cambrian information revolution had in bringing about very “modern” levels of information processing.

Though we focus on the large share of brain-bearing fauna, two other patterns for the other nervous system levels are worth mentioning—the decrease in the neuron-less category (sponges) post-Cambrian and the rise in the ganglionated one. Sponges, previously noted to be among the largest components of BST faunas (Botting and Peel 2016), are considered the dominant Cambrian sessile suspension feeders (Wu et al. 2014), and their proportional importance may have dropped as other animals radiated, including those taking similar niches. For instance, they are present but a minor component of the ecologically diverse and well-sampled living Bermuda fauna in our data. However, a lack of sponges in some Lagerstätten may also reflect difficulty of preservation, as they often fall apart as isolated spicules. Some of the BST Cambrian Lagerstätten may be better at capturing them intact. The rise of ganglionated taxa heavily reflects the radiation of bivalves and bryozoans, which were absent or rare in the Cambrian faunas, though we also consider some potential underrepresentation of ganglionated Cambrian invertebrates among the excluded “problematica” based on uncertain coding (well-cephalized groups like arthropods and decentralized ones such as echinoderms were perhaps more clearly assigned to their nervous system level). A Phanerozoic increase in ganglionated animals relative to those with simpler nervous systems might also reflect an increase in motile benthos able to respond to disturbance (Bush et al. 2007; Manojlovic and Clapham 2021), a change expected to increase at least some information-processing abilities.

As with any ecological analysis performed with fossil data, there is always the consideration that taphonomy may obscure the original biological trends of interest. Though they include many soft-bodied or poorly mineralized taxa and allow much more complete ecosystem reconstructions than most other fossil assemblages, the Lagerstätten we examined would still miss many elements of the in-life fauna or have them altered by decay and transport (Caron and Jackson 2006; Nanglu et al. 2015). For instance, although BST deposits contain exceptionally preserved priapulids and polychaetes, they lack the even softer nemerteans or flatworms (Butterfield 2003), which are recorded in the modern Bermuda faunal list. The sporadic distribution of Lagerstätten across time and space, as well as the potential unusualness of the paleoenvironments they represent, has frustrated their use in ecological or evolutionary analyses (Butterfield 1995, 2003; Gaines 2014). In addition, different Lagerstätten can have rather different preservational biases (Allison 1988; Saleh et al. 2020), adding noise to our analyses. Nonetheless, despite the great variability in (paleo)environments, preservational mode and history of study for our 23 faunal assemblages across the Phanerozoic, it is noteworthy that the Cambrian and post-Cambrian faunas are so similar on average in the sensory and cognitive attributes we examined. Likewise, the modern Bermuda fauna was sampled and studied in a completely different manner, yet it displays reasonably similar statistics to the average fossil assemblage. This suggests a major underlying macroecological pattern could be present. Although we only analyzed one modern faunal list, modern biodiversity censusing of the world's oceans has shown that three groups well represented in sensory and cognitive complexity—crustaceans, mollusks, and fish—are estimated to make up close to half the living species in most geographic regions (Costello et al. 2010).

Overall, our results underscore the importance of the Cambrian information revolution. The change in share of animals with complicated sensory and nervous systems between the Cambrian and post-Cambrian is minor compared with the relatively rapid evolution and development of information-processing systems in animals during the late Ediacaran–Cambrian transition. Fossils in the Ediacaran period have so far lacked macroscopic sense organs (Marshall 2006), and evidence for cephalization or brains is uncertain (Northcutt 2012), in stark contrast with all Phanerozoic ecosystems. This does not mean, however, that Ediacaran animals completely lacked sensory and nervous systems before the Cambrian information revolution. Evidence in the late Ediacaran of possible trackways of bilaterians walking with paired appendages (Chen et al. 2018), selective drilling by unknown predators (Hua et al. 2003), as well as the mollusk-like Kimberella and its feeding and locomotion traces (Gehling et al. 2014; Ivantsov et al. 2019) attest to at least rudimentary information-processing systems. Although the complexity of these systems remains unknown, they were sufficient to seek resources with directed locomotion and to manipulate food.

Due to the uncertainties involved, we did not attempt to quantify the neurological complexity of Ediacaran faunas in the same way as Phanerozoic faunas, though as previously mentioned, it is likely that at least diffuse nerve nets existed (Northcutt 2012). In any case, the major development of bilaterian body plans that include recognizable sense organs and brains did not arise occur until the Cambrian radiation, even if these innovations relied on developmental tool kits that existed earlier in the Neoproterozoic (Erwin 2020). Although our Lagerstätte-based analysis begins temporally with the Chengjiang and Sirius Passet faunas in Cambrian Stage 3, complex sense organ–bearing and brain-bearing taxa were expanding earlier in the Cambrian, as the trace fossil record attests (Carbone and Narbonne 2014; Mángano and Buatois 2020). Specifically, the continuous trace fossil record across this interval records the behaviors of sensorially and cognitively complex bilaterians, including euarthropods, that would come to dominate the body fossil assemblages sampled in our analyses (Mángano and Buatois 2020).

All in all, evidence suggests that new ways of sensing, processing, and responding to information developed early in marine ecosystems in the Cambrian information revolution, as anatomical data reveal their already widespread presence by Cambrian Stage 3. Subsequent to the Cambrian, more sophisticated variants of these systems evolved, but change was a matter of degree rather than kind (e.g., there were increases in the acuity of eyes or processing power of brains in many lineages). This is somewhat analogous to how most phyla and body plans were present by the Cambrian, with diversification and elaboration happening within them later. We can also compare our results with Aberhan et al.'s (2012) previous findings that the proportional diversity of marine invertebrate genera possessing vision peaked early in the Phanerozoic, continuing at a lower, stable level after the Ordovician. Because we included chordates, we find the share of eyes in marine assemblages to somewhat increase post-Cambrian (excluding chordates, the increase in the share is smaller, with an 11% rather than an 18% difference in proportion between Cambrian and post-Cambrian, and the two intervals would not significantly differ statistically by our resampling test). Our results also differ from those of Aberhan et al. (2012) in finding overall higher shares of eye-bearing genera across the Phanerozoic, possibly because we examined data from select Lagerstätten and a modern assemblage, where they might be more easily recorded than in most fossil assemblages. No or little rise in the average post-Cambrian share of the fauna with macroscopic sense organs and brains might represent “saturation” of ecospace or of life modes where sensory and cognitive complexity is required, consistent with Bush et al.'s (2011) findings that much of modern ecospace in terms of tiering, motility, and feeding modes was already occupied by the Cambrian. Similarly, Dunne et al. (2008) showed that Cambrian food webs are very similar to modern ones. There may be only so many niches in an ecosystem that require complex cognition, at least in proportional terms. This would be consistent with such a similar average share of genera (around two-thirds) possessing brains in our analysis for both the Cambrian and post-Cambrian in spite of such a great shift in the balance of which taxonomic groups possessed them through the Phanerozoic. Similarly, the largest rise in diversity of trace fossils occurs during the Cambrian radiation, with later increases more modest or gradual, as found by Buatois and Mángano (2018).

Future research could also examine changes in the complexity of macroscopic sense organs or brains over the Phanerozoic, in addition to their presence or absence. For instance, the neural architectures seen in Cambrian panarthropods (Ma et al. 2012; Tanaka et al. 2013; Cong et al. 2014; Strausfeld 2015) are quite conserved and similar in complexity to those in the present. Likewise, Ma et al. (2012: p. 258) characterize compound eyes in the Cambrian as being “in size and resolution, equal to those of modern insects and malacostracans.” Such inference about Cambrian compound eyes is seen, for instance, from Paterson et al.'s (2011, 2020) finding that the stem-arthropod Anomalocaris compared well to most living arthropods in eye size, high ommatidial lens count per eye, and low interommatidial angles, all factors that correlate with visual acuity in the present. These claims would imply that modern levels of cognition and sensory acuity were achieved exceptionally early for arthropods. However, the situation may be different for chordates or cephalopods, with many large-brained and behaviorally sophisticated examples only appearing or diversifying later in the Mesozoic and Cenozoic (e.g., teleosts, cetaceans, and coleoids). These animals probably have far more sophisticated nervous systems than their Cambrian ancestors, and they typically occupy more active life modes; small-brained Cambrian chordates were suspension feeders but not yet predators, and Cambrian cephalopods were slowly motile rather than fast motile. In contrast, Cambrian arthropods already occupied both fast and predatory life modes (based on data from Bambach et al. 2007).

Living non-vertebrate chordates (urochordates and cephalochordates) are reported to have only 10²–10⁴ neurons in their brains, whereas vertebrates have 10⁷–10¹⁴ (Meinertzhagen 2010). (For context, annelid brains have 10³–10⁴ neurons; arthropod brains, 10⁵–10⁶; and mollusk brains, 10³–10⁸; with gastropods occupying the lower end of the range and cephalopods the upper end.) There is a long history of interest in the idea of a directional trend in cognitive evolution for vertebrates, although much of the literature pertains to terrestrial tetrapods. Russell (1981, 1983) noted that for animals (mostly vertebrates), average maximum encephalization, or proportional brain mass, increased through the Phanerozoic. He suggested that this was an example of “exponential evolution” or acceleration in biological complexity. Jerison (1970, 1973) also chronicled a gradual, progressive brain-size rise through the Mesozoic and Cenozoic among vertebrates.

Even if the proportion of animals with brains did not trend appreciably through the Phanerozoic, it is possible that there are more neurons firing in modern marine ecosystems than in Cambrian ones, which lacked big-brained vertebrates and cephalopods. One might expect that the prevalence of sensory and nervous systems is similar over time in similar niches, but with systems becoming more finely honed by pressures like predator–prey escalation. However, in some cases, novel sensory modes such as echolocation in cetaceans arose later in the Phanerozoic. Future research might examine temporal trends in the presence and prevalence of these later-evolving systems, in addition to the two types of macroscopic sense organs examined here. It is worth mentioning, too, that our ecospace analysis focused on feeding mechanism, habitat, and motility, as is commonly done in paleontology; this focus might underweight the importance of many other biological functions such as communication, social signaling, mating, and reproduction (Bush et al. 2016), for which nervous system complexity also plays a major role but which are more difficult to study from fossil data (Hsieh and Plotnick 2020). These behaviors are nonetheless highly relevant in the evolution of information processing in animals, and they may have helped promote diversification (Bush et al. 2016).

Though Cambrian and post-Cambrian marine faunas are quite similar by our metrics, it is worth noting that terrestrial ecosystems are even more overwhelmingly dominated by brains. Only a subset of metazoans became truly terrestrial (Selden 2016), with three of the major taxa—tetrapods, panarthropods (including onychophora), and gastropods—possessing brains. Other taxa with terrestrial members, such as oligochaete annelids and platyhelminth flatworms, at least possess ganglia. No animals with decentralized nervous systems, or lacking nervous systems, made it onto land, although some colonized continental aquatic environments. The majority of animal species alive today have the cognition-associated anatomical traits we looked at in our analyses—brains, eyes, and chemoreceptive organs—due to the overwhelming richness of terrestrial insects, whose diversification has been the topic of much study (Labandeira and Sepkoski 1993; Vermeij and Grosberg 2010; Clapham et al. 2016).

With some exceptions, such as marine mammals, most well-studied directional increases in brain size have also been in terrestrial vertebrates. For instance, Cenozoic birds were larger brained than their Mesozoic counterparts according to Milner and Walsh (2009). Jerison (1970) similarly argued for a rise in relative mammalian brain size alongside an increase in overall range, though this was disputed by Radinsky (1978). A land-based lifestyle might strongly select for, if not outright require, complex active bodies with embodied cognition (as described by Trestman 2013), as many passive, immobile lifestyles are not viable for animals due to numerous differences between living aquatically and terrestrially (Denny 1993; Vermeij and Dudley 2000; Grosberg et al. 2012; Burgess et al. 2016; Vermeij 2017). For example, marine animals have the option of either moving toward a desired resource or waiting for it to be transported to them (Bambach et al. 2007), but the inability of air currents to carry solid mass would heavily promote the former strategy and any navigational abilities it may entail. In many cases, terrestrial habitats may also be spatially more heterogeneous and complex (Grosberg et al. 2012), further favoring increased cognition (Mugan and MacIver 2020). Future research should explore the sensory, cognitive, and behavioral aspects of the water-to-land transition (e.g., Sturmbauer et al. 1996; Ord and Hsieh 2011; Ladich and Winkler 2017; Mugan and MacIver 2020).

Finally, although we examined and compared cognition in faunas using taxonomic diversity, it is worth considering a perspective from abundance. Diversity and abundance are frequently, but not always, correlated (Clapham et al. 2006), though accurate abundance data for fossil communities can often be difficult to obtain. Our view based on presence or absence of animal genera with cognition-related traits in each assemblage may miss aspects of ecological function only inferable through either absolute numbers of individuals or their relative proportions within an ecosystem (Vermeij and Herbert 2004; Jackson and Erwin 2006; Turvey and Blackburn 2011; Winfree et al. 2015; Gaston et al. 2018). Conservatively, our data tell us at least one individual animal with the given sensory or cognitive trait was present at a site and, if not transported there as bodily remains after death, had been interacting with the environment and processing information there in life (additionally, given how mobile many cognitively complex taxa are, mere presence at a fossil locality at a single point in time may not record the degree or intensity of interaction with the local environment—some long-distance pelagic migratory species may pass through the location briefly, while others may spend much greater time sensing and reacting to the habitat and its residents). Our presence-based analysis would thus not say how relatively “important” an animal was in the ecosystem in, for example, commonness of encounter or interaction with other individuals of the same or different genera.

Well-studied and thoroughly sampled community data from the Burgess Shale and Chengjiang biotas show arthropods as a group dominating not just taxonomic diversity but also in number of individuals (Caron and Jackson 2008; Zhao et al. 2014; Nanglu et al. 2020) in Cambrian ecosystems. Even within the best-preserved and best-sampled Lagerstätten, we acknowledge that taphonomy may still influence this view, for arthropods have exoskeletons that strengthen preservation relative to soft-bodied or tiny animals that may be also quite abundant in raw numbers in marine ecosystems. In the present day, the preponderance of the world's animal biomass is estimated to be in arthropods (though nematodes lead by numerical abundance; Bar-On et al. 2018; Van Den Hoogen et al. 2019). Bar-On et al. (2018) estimated that of the roughly 2 gigatons of carbon (Gt C) total of contemporary animal biomass, arthropods make up ≈ 1 Gt C, with fish at ≈ 0.7 Gt C and mollusks and annelids at around 0.2 Gt C. Though still a small share of all metazoan biomass, humans (≈ 0.06 Gt C) and their livestock (≈ 0.1 Gt C) now surpass all other non-fish vertebrates, having done so in a relatively geologically short span of time. In any case, an overwhelming share of animal biomass today is concentrated in bodies controlled with brains and sensory systems. The Cambrian information revolution fostered not only the diversification of cognitively complex organisms, but also their incredible abundance and biomass from the beginnings of the Phanerozoic up to today.

Conclusion

Comparisons of faunal lists from Cambrian and post-Cambrian ecosystems reveal similarly high shares of animal genera with brains as well as macroscopic sensory organs. Our results show that the Cambrian radiation generated ecosystems that were very “modern” in sensory- and information-processing complexity, comparable to many ecosystems later in the Phanerozoic, even if they were less diverse and occupied less total ecospace. This is consistent with behavioral evidence, including trace fossils, showing “modern” types of behavior existed by the Cambrian and the fact that most of the body plans and life modes requiring complex information processing (e.g., mobile predators) were present then. A major difference, however, is that the overwhelming majority of sensorially and cognitively complex animals in the Cambrian were panarthropods, whereas chordates and mollusks further diversified in later times. In both Cambrian and modern times, nervous systems permitted a variety of life modes, but the most active modes (those categorized as fully motile fast by Bambach et al. [2007]) are almost exclusively associated with brains, which first appear in the Cambrian. The increase in information-processing abilities in the Cambrian is exceptional in the history of life, equaled only perhaps during the conquest of the land. The Cambrian is likely the earliest time when a cognitive or behavioral biologist would find lots of interesting things to study.

The name “Phanerozoic” alludes to a time of macroscopic life, revealed as abundant fossils visible to the naked eye (Chadwick 1930; Schopf 1994). From the Greek for “manifest” or “visible,” the term “phaneron” had also been used in philosophy to describe all that we can observe or perceive through our faculties, the “collective total of all that is in any way or in any sense present to the mind” as defined by C. S. Peirce (Hartshorne and Weis 1931: p. 141). With the Cambrian information revolution ushering in a host of creatures that sensed and perceived the world around them, the moniker of this eon is apropos in more ways than one.