Invertebrate Sentience: Summary of findings, Part 2

By Daniela Waldhorn @ 2019-06-14T21:19 (+71)

Executive Summary

Invertebrate welfare has been gaining recent traction in the scientific literature and among the effective altruism community. However, whether invertebrates have the capacity to experience pain and pleasure in a morally significant way is still very uncertain. Given this current epistemic state, Rethink Priorities has been compiling and analyzing relevant scientific evidence regarding invertebrate consciousness and exploring criteria that evaluate whether individuals of a given invertebrate species or taxon have valenced experience. These data have been gathered and displayed in this database. In a prior post, we summarized our main findings by feature. Here, we present our results by taxa.

Which consciousness-indicating features did we find in invertebrates? Results by taxa

Invertebrates comprise an enormous and diverse array of animals, from simple organisms, such as C. elegans, to more complex individuals, like octopuses. They live in a variety of aquatic and terrestrial environments, and individuals of extant species span an enormous range of body organizations –for instance, their nervous systems may range from a complex centralized system to simple nerve nets.

Hence, it must be conceded that the numerous invertebrate species and their diversity imposes severe constraints to justifiable generalizations about the presence of consciousness in this group of animals. It also poses an important limitation regarding the invertebrate categories included in our study: to which extent our findings are generalizable for an entire taxon of animals?

As addressed in a previous post, the decision about the taxonomic level at which we would investigate the selected animals was guided by a balance of two considerations: (i) choosing a high taxonomic rank that, still, would provide informative evidence, and (ii) the amount of existing research which has been conducted at each taxonomic rank. The above criteria led us to consider varying taxonomic ranks –in some cases an invertebrate category covers a fairly narrow taxon (genus or species), but in other cases, it comprises a much larger group of animals (order or infraorder). This is shown in the table below:

Table 1. Studied organism categories and the number of species that they comprise.^[1]

When we were pushed to consider relatively high taxonomic ranks, a single organism category comprises numerous –sometimes, thousands of– species. In these cases (e.g. order Araneae), generalizing from a single species to the rest of the taxon is potentially problematic. However, this methodological decision was mostly founded on the amount of current research for the taxon in question.

Furthermore, the extent to which a taxon has been investigated plays another crucial role throughout all of our findings: broadly, we observe that the presence of consciousness-indicating features is directly conditioned by the extent to which a species or taxon has been investigated. That is, we have more evidence that certain animals are conscious to the extent that they have been the subject of scientific research. There is a significant correlation (-0.86) between the percentage of "unknown" responses and the degree of features present per species/taxon. However, there are two important exceptions for this claim: C. elegans and Aplysia. These animals have been widely used as research models, but not such strong evidence arises about whether they are conscious or not.

There is greater evidence about the presence of consciousness-indicating features in fruit flies, honey bees, octopuses, crayfish and crabs. These are, at the same time, the categories for which there are the most positive responses. Nevertheless, that does not necessarily entail that these animals are “more conscious” than other invertebrates. It only points out that we know more about consciousness in these animals than in individuals of other taxa.

In contrast, we found the greatest gaps in knowledge for moon jellyfish, spiders, and earthworms. These results are presented in the graph below (graph 1). Invertebrate species appear in blue, vertebrate species in green:

Graph 1. Lack of research about consciousness-indicating features according to available evidence.

Our main findings for each of the revised invertebrate categories are presented hereunder. We describe first the results for those species or taxa regarding which a higher percentage of positive evidence was found.

Fruit flies

A good example of the relationship between the presence of consciousness-indicating features and the extent to which a species has been investigated are fruit flies (Drosophila melanogaster). Fruit flies have been widely used as biological models and they feature widely in studies of invertebrate nociception. Hence, our findings about fruit fly sentience are more solid than evidence regarding any other taxon we studied—we found direct (positive or negative) evidence for 73.4% of the features. Consistently, the lack of research in fruit flies is the lowest among all the invertebrate species reviewed—“unknown” responses account for 20.4%. Moreover, we found positive evidence (“likely yes” and “lean yes”) for 71.4% of the features. In contrast, we did not locate direct negative evidence against any of the features. Yet other research suggests that four features (8.1%) have not been observed in these insects or that there is a low chance that it is found in them.

Existing literature points out that fruit flies have complex and centralized nervous systems, with specialized modulatory neuron clusters that modify neural activity according to the physiological and motivational state of the insect. Other structures integrate mechanosensory and proprioceptive information (Sareen et al., 2011; van Swinderen, 2005; van Swinderen & Greenspan, 2003). This and other findings suggest the presence of centralized information processing in fruit flies. Furthermore, the neurons used by fruit flies to detect noxious stimuli have been precisely mapped (Al-Anzi et al., 2006; Aldrich et al., 2010; Guo et al., 2014; Hwang, 2015; Im & Galko, 2012; Turner et al., 2016).

Consistent with these anatomical characteristics, fruit flies display various noxious stimuli reactions, such as cardiovascular and other physiological responses to mechanical constraints (Paternostro et al., 2001; Sénatore et al., 2010). They display active defensive behaviors when attacked or when competing with other flies for mating opportunities, territory, and dominance (Chen et al., 2002; Kim et al., 2018; Li et al., 2016). When fruit flies are coated in dust, for instance, they show what is probably a protective behavior—they groom themselves (Corfas & Dudai, 1989; Kays et al., 2014; Seeds et al., 2014). Nevertheless, it must be considered that the latter is a highly stereotyped reaction.

Moreover, overhead shadows cause fruit flies to disperse, a response aimed at avoiding potential predators. However, if the shadow overpasses starved fruit flies, they will continue feeding (Gibson et al., 2015). This tradeoff behavior suggests that flies may possess some kind of unified utility function in which benefits and risks of a given situation are processed and evaluated.

In one of the first scientific experiments on Drosophila learning and nociceptive reactions, fruit flies were trained with sequential presentations of an odor (conditional stimulus) and electric shock (unconditional stimulus). The odor predicted shock and fruit flies subsequently avoided it. The memory of this association persisted for 24 hours (Quinn et al., 1974)—which is between 3.3% and 4% of the average lifespan of a fly. In a more recent experiment, it was demonstrated that fruit flies could even display ‘relief’ learning: in this case, the odor was presented after the shock. Fruit flies learned that the aroma predicted relief from the noxious stimulus, and consequently, approached it (Yarali et al., 2008).

Are these experiments evidence of consciousness or are the fruit flies’ responses driven by mere non-conscious associations? We do not know with certainty. However, further research has shown that fruit flies are also capable of other more complex forms of learning, such as operant conditioning and contextual learning. For instance, in an experiment, flies learned that in one context a specific color predicted punishment while in another environment it was a sign of reward (Brembs & Wiener, 2006). They probably learn from observing other peers’ behavior (see e.g. Sarin & Dukas, 2009), and they learn about new food locations, demonstrating that they also possess spatial memory—among other navigational skills (Ofstad et al., 2011).

Remarkably, there is partial evidence for some mood state behaviors in fruit flies. In particular, anhedonia-like behavior can be induced in fruit flies by exposing them to aversive, uncontrollable vibrations over several days (Ries et al., 2017). Better known are their learned helplessness responses, which were first investigated in 1996. In that experiment, fruit flies were exposed to inescapable mechanical shaking in a black-white Y-maze escape task. Later, in a shuttle box escape task, those fruit flies had longer escape latencies than other flies that had been subjected to escapable shaking or no shaking (Brown et al., 1996). By the same token, in stressful situations, it has been found that fruit flies exhibit physiological and behavioral responses that are strikingly similar to an anxiety-like state—as it had been previously observed in rodents (Mohammad et al., 2016).

In turn, fruit flies also seem to have positive valenced experiences, and seem to be willing to pay a cost in order to get a reward. In a study, fruit flies endured electric shocks in order to attaining the cue associated with ethanol, trading off the punishment for the drug (Kaun et al., 2011).

Recently, opioid-like peptides have been discovered in fruit flies (Santoro et al., 1990; see also Harrison et al., 1994 and Sneddon, 2017). Additionally, they are affected by analgesics, antidepressants or anxiolytics, as well as by recreational drugs in a manner similar to humans. Drugs like cocaine or alcohol, for example, induce motor behaviors in fruit flies that are strikingly similar to those induced in vertebrates (Hirsh, 2001). Moreover, chronic alcohol drinking in fruit flies leads to increase alcohol consumption, as seen in humans who show alcohol addiction (Devineni & Heberlein, 2009 in Scholz & Mustard, 2011).

Honey bees

After fruit flies, we found that honey bees (Apis mellifera) display a considerable amount of consciousness-indicating features. We encountered positive evidence (“likely yes” and “lean yes”) for 63.3% of the reviewed indicators. However, we did find negative evidence for one of the studied features: in particular, when honey bees are injured, they do not self-administer an analgesic at their disposal (explained below). Additional research suggests that the other 6% of the features may not be observed or that there is a low chance that they are found in these insects. Lastly, the lack of evidence is higher than that for flies, accounting for 30.6% of the responses.

Notably, some of the the higher-order features of learning seen in vertebrates can be found in honey bees, such as conditioning of relatively complex actions and social learning. For example, bees can be conditioned to various operational actions, such as entering a hole to obtain food (Sokolowskia & Abramson, 2009), or touching a silver plate with their antennae to get sucrose (Kisch & Haupt, 2009). However, what is more surprising in honey bees are their cognitive skills, which allow them, for instance, to use a symbolic ‘language’ (the honeybee dance) for transferring route information to other bees. Furthermore, it has also been reported that honey bees imitate bumble bees' techniques of nectar-robbing by cutting holes into flower spurs and extracting nectar without pollinating the flowers, a form of observational learning (Leadbeater & Chittka, 2007).

We did not find any specific investigation about tool use in honey bees. However, the ability of honey bees to learn through operant conditioning (Erber et al., 2000; Kirchner et al., 1991; Kisch & Erber, 1999; Kisch & Haupt, 2009) and other research with bumble bees^[2], strongly advise for this skill to be studied in honey bees. Bumble bees, in particular, are able to learn to pull a string to reach an artificial flower containing a sugar solution (Alem et al., 2016; see the video), and even to “play golf”. In an experiment, bumble bees learned to move a ball into a hole, especially when they were shown how to do it by a plastic bee or a bee that had been previously trained to do the task (Loukola et al., 2017; see the video).

Honey bees can also solve context-dependent problems. In an experiment, bees learned that when a certain color was presented to them, one odor predicted sucrose reward and another predicted no reward. However, when a different color was presented, these relationships switched (Mota et al., 2011). In another study, honey bees learned that artificial flowers of a certain color were rewarding on certain times of the day and that they had to approach these flowers from a certain angle in order to obtain reward—land on a specific petal first (Gould, 1987).

There is partial evidence that honey bees could be aware of their certainty or uncertainty when making a choice. This form of metacognition (uncertainty monitoring) was explored in an experiment where bees had to respond to a visual discrimination task that varied in difficulty between trials. Bees were rewarded for a correct choice and punished for an incorrect one. When given the choice to avoid choosing by exiting the trial, honey bees did so. They opted out more often on difficult trials, and opting out improved their proportion of successful trials (Perry & Barron, 2013).

Notwithstanding the above, when injured, honey bees do not show a greater preference towards a morphine-based analgesic solution, as it might have been expected. Injured bees do increase their food intake, probably due to an increased energetic demand for an immune response (Groening et al., 2017). But when exposed to a different noxious stimulus—i.e. an electroshock—morphine appears to relieve pain in honeybees (Nuñez et al., 1983). Thus, it seems that certain reactions to noxious stimuli vary not only among different species, but also with the stimulus.

Despite our extensive knowledge of the complexity of honey bee behavior, it is not known whether bees possess nociceptors or other equivalent structures. Nevertheless, it has been demonstrated that their behavioral complexity requires integration in a central structure, facilitated by specific neural mechanisms found in their central nervous system (Menzel & Giurfa, 2001; Paulk et al., 2014). Moreover, honey bees’ navigational skills—including spatial memory—are functions that in humans and in vertebrates have been typically associated to the midbrain. Indeed, the neurons and the circuit involved in path integration memory in honey bees have been already described (Stone et al., 2017).

Despite certain limitations (in particular, on the bee's pain system), the state of current knowledge for the different features in bees—including those referred to their abilities for solving novel challenges—and their high level of social organization^[3], suggest that these insects are possibly sentient.

Octopuses

The impact of additional research in our findings about invertebrate consciousness is clearly exemplified in the case of cephalopods and decapod crustaceans. Recent research has shown that these animals are highly intelligent and potentially capable of experiencing pain. This has led to the inclusion of cephalopods in animal protection legislation in some jurisdictions (e.g. in the European Union, see EFSA, 2005). Consistent with these findings, we have observed evidence that octopuses (family Octopodidae), crayfish (family Cambaridae) and crabs (infraorder Brachyura) perform various behaviors suggestive of painful experiences.

There is an important body of research pointing out that octopuses are conscious, capable and smart individuals. Specifically, regarding the features we studied, we found positive evidence (“likely yes” and “lean yes”) for 63.2% of the cases. In contrast, direct and indirect evidence point out that 8% of the reviewed features have not been observed or that there are low chances to find them in these animals. The potential consciousness indicators regarding which we did not find empirical research amount to 28.6%.

Of known invertebrates, octopuses have the most neurons, notably complex brains and the most complex central nervous system, with a clear hierarchical organisation. Moreover, recent neurophysiological studies have provided direct evidence for the presence of nociceptors (Di Cristina, 2017; Fiorito et al., 2015). These structures are likely to be connected by nervous pathways to the ‘lower’ parts of their nervous system (Fiorito et al., 2015). Consistently, various noxious stimuli reactions have been documented in octopuses. Their arms, for instance, are capable of reflex withdrawal to a noxious stimulus (Hague et al., 2013). They jet away when threatened by another animal of before a predator (Anderson et al., 2010). When attacked, octopuses use their arms to defend themselves (Wells, 1978). They can expulse ink or bite a predator or whatever is molesting them (Anderson et al., 2010; Caldwell, 2005). Physiological responses to nociception have also been observed in these animals (Andrews et al., 2013; Della Rocca et al., 2015; Ross, 1971).

Octopuses are exploratory and skilled navigators. For instance, they will make detours to get a prey seen through a transparent barrier, even though the detours carry them temporarily out of sight of the prey. The possibility of nonvisual intermediate cues cannot be ruled out experimentally (Wells, 1964). Octopuses record and remember information about their environment and spatial orientation. In the literature, they are described as “adept at exploratory learning”. They use single visual cues, code the location of a goal in space with a nearby landmark, and appear to use multiple beacons flexibly during a single navigation task (Grasso & Basil, 2009; Mather, 1991).

Octopuses show sensitization, habituation, associative learning, including visual and tactile discriminative capabilities, and spatial learning (Hochner et al., 2006). Moreover, octopuses learn to perform unfamiliar actions, such as inserting an arm up a tube and out of the water in order to obtain food (Crancher et al., 1972). They can learn how to open a jar, but even more interestingly, they learn faster how to do it from watching and imitating other octopuses that succeed in this task (observational learning, Fiorito et al., 2009).

Furthermore, it can be argued that octopuses display a degree of self-control: in an experiment, it was observed that they avoid attacking a prey (hermit crabs) if they have stinging sea anemones attached to their shells. This avoidance behavior was observed even 24 hours later (Ross, 1971). Furthermore, they adjust foraging according to predation risk: if the latter is high, octopuses maximize efficiency not by optimizing energy gain but by avoiding the risk of injury or death to ensure survival to reproduction (Mather & O'Dor, 1991).

Octopuses use tools as well. Soft-sediment dwelling octopuses (Amphioctopus marginatus), for example, retrieve coconut shell halves discarded by the local human population and later assemble the shell halves into protective shelters. The awkward manner in which the octopuses must move while carrying these shells (the authors describe it as “stilt-walking”) represents a cost in terms of energy and increased predation risk, which is only recouped later when the shelves are successfully assembled into a surface shelter or encapsulating lair. Importantly, the only known source of these clean and lightweight shells is the coastal human communities, and thus the octopuses have not interacted with these items on an evolutionary timescale (Finn et al., 2009). Still, some researchers argue that little is known about other potential cognitive abilities in octopuses (Richter et al., 2016).

Octopuses are the only invertebrates in which we found playing behavior, an activity best known in vertebrates, especially mammals. In an experiment with common octopuses (Octopus Vulgaris), the animals were given two Lego objects during various consecutive days. After exploring the pieces, most of the octopuses started pulling the Lego blocks closer or pushing them away, time and again, in a coherent action. They also towed the floating Lego blocks on the water surface, in a manner that qualifies as a play-like behavior (Kuba et al., 2006). Catherine Cservenka (2015), who looks after Ollie –a giant Pacific octopus– at a sea life park in Weymouth, England, explains that Ollie also likes to play with toys of different shapes and textures. Although Ollie belongs to a different octopuses’ family than the one we considered (Enteroctopodidae), his favorite toys are blocks of Lego as well. Ollie likes to hold the plastic building blocks with his tentacles, although his keepers admit he has yet to build anything recognisable.

Additionally, there is evidence that octopuses—when transported or exposed to repeated noxious stimuli—may experience distress. In those circumstances, they spend an extended amount of time with the arms curled over the body (a defensive posture) (Fiorito et al., 2015), begin to writhe, change their coloration, stiffen the whole body and/or expel intramantle ink—which is toxic to the octopuses themselves. These, as well as displacement behaviors, have also been observed in confined octopuses. For example, in those held in laboratories and used for investigations (Bennett & Toll, 2011; Budelmann, 1995). However, research about mood state behaviors in octopuses is still very limited.

Crayfish and crabs

We know much less about crabs and crayfish than fruit flies or octopuses. Still and as recent research points out, there is important evidence regarding the possibility that these crustaceans are conscious.

Crabs have fewer neurons than honey bees and ants, yet current evidence suggests that crabs probably display more than half of the features we studied (57.2% of responses are “lean yes” and “likely yes”), although there is no evidence regarding 36.7% of them. We did not find any direct negative evidence for crabs, albeit there is a lesser likelihood for the presence of one of the features (flexible tool use), which constitutes 2% of the indicators.

Crabs’ nervous system appears to centralize information processing (Brusca & Brusca, 2003; Sømme, 2005; Zeil & Hemmi, 2014), but it must be considered that this function comes in degrees. Similarly, different ganglia of the central nervous system process spatial information and organization of movement (see Brusca & Brusca, 2003; Sømme, 2005). Besides, opioid-like receptors have been described in these crustaceans as well (see e.g. Henke et al., 1997), and in various circumstances, they are affected by analgesics like morphine in a manner similar to humans (Barr et al., 2008; Elwood et al., 2009; Lozada et al., 1998; Maldonado et al., 1989). However, no studies have yet identified nociceptors or receptive fields in decapods (Sneddon, 2017).

Still, there is evidence that crabs might be conscious. Various studies indicate that crabs show physiological responses induced by nociceptive stimulus or handling (Elwood et al., 2009; Patterson et al., 2007; Dyuizen et al., 2012) and they move away or escape when threatened—for example, when facing a predator (Robinson et al., 1970; Zimmer-Faust et al., 1994). If a claw of a Cancer pagurus crab is removed by pulling it off, the crab will repeatedly touch the wound with its appendages (McCambridge, personal communication in Sneddon et al., 2014). Further, formalin injection into one claw of shore crabs, H. sanguineus, induces shaking and rubbing of the appendage and the use of that appendage is markedly reduced (Dyuizen et al., 2012). Besides these behaviors, it has also been observed that crabs react defensively when exposed to noxious stimuli. For example, in this experiment crabs (Chasmagnathus granulatus) engage in a defensive threat display right after being electrically shocked (Lozada et al., 1988). Boxing crabs even use sea anemones as tools for defending themselves (Guinot et al., 1995)^[4].

Crabs also show sensitization, habituation, and can learn through classical and operant conditioning. In fact, they can learn unfamiliar actions such as pressing a lever to obtain food (Abramson & Feinman, 1990). There is also evidence that they respond to a stimulus according to the context and have contextual associative memory (Fustiñana et al., 2012).

Notably, crabs cannot only learn to avoid noxious stimuli—they can even make tradeoffs and remember their learned avoidance for at least 24 hours. In a study, ninety crabs (Carcinus maenas) were individually placed in a brightly lit arena, and had the option of scuttling to two dark shelters. First, it must be considered that crabs, in general, prefer taking shelter in dark and hidden places, in order to avoid being spotted and eaten by predators. This strategy guarantees safety but is also costly due to a trade-off between hiding time and time spent in other essential activities, like feeding, reproduction or defending resources (Guerra-Bobo & Brough, 2011; see also Mosknes et al., 2003; Tapia-Lewin & Pardo, 2014). In the experiment, once the crabs had taken refuge away from the light, half were given an electric shock in the first shelter they chose. Then, the shocked crabs were placed back into the tank again. Most of them moved back to the original shelter where they had been stunned. Those that made this decision were then shocked a second time. Now, the second electric shock produced a significant switch in the crabs’ behavior. They were placed back in the arena another eight times, and although there were no more shocks, they continued to avoid the shelter where they had been sparked (Magee & Elwood, 2013).

Thus, crabs are willing to pay a cost—leave the dark shelter, a desired place, and go out into a potentially dangerous lit environment—in order to avoid a noxious stimulus. In a similar study, it was observed that the probability of a crab moving out of the shelter was increased if the external light intensity was low. That is, the response to the electric shock was modulated by the motivation to avoid brightly lit areas (Elwood & Mansoor, unpublished in Elwood et al., 2009). This response can be seen as a rudimentary expression of self-control—or, at least, crab’s behavior cannot be explained as a simple reflex reaction. The shocked crabs learned from their experience and this was driving their future choices. Trade-offs, as those seen in crabs, clearly involve some form of processing in which the different needs of the individual are weighed. In the researchers’ words, these observations fit a “key criterion of pain” for these animals (Magee & Elwood, 2013).

Much less is known about other features related to cognitive sophistication and mood state behaviors in crabs. However, according to the literature, crabs probably experience stress in ways that are analogous to those of vertebrates (Elwood et al., 2009). One anxiety-like behavior in crabs is reluctance to explore. In an experiment, this behavior was induced after exposing the crabs to a luminous chamber. Later, this reaction was reversed with the administration of the antidepressant fluoxetine—in humans, prescribed for the treatment of a variety of psychological disorders including depression and anxiety (Gallup & Tresguerres, 2016). Although this study suggests that crabs may experience a form of anxiety, further research is needed about emotional states in these animals.

Crayfish, for their part, also show more than half of the considered features (54% of the responses are “lean yes” and “likely yes”), although there is no evidence for around a third of them (34%). We did not find any direct negative evidence for these crustaceans, albeit there is indirect evidence that suggests that six features (8%) are not observed or that there are low chances that they are found in crayfish.

When compared to octopuses and crabs, crayfish seem to have a less sophisticated nervous system. Their chain of segmental ganglia and its functioning suggest that their information processing is less centralized than in crabs. Nevertheless, these animals respond quickly and strongly to high temperatures, which are considered to be a potentially relevant noxious stimulus for crayfish. High temperatures can be detected by sensory neurons, which may be specialized nociceptors in crayfish. However, these animals show no response to low temperature stimuli and to specific chemicals that do not damage tissue (Puri & Faulkes, 2015).

How do crayfish react when touched with a very high temperature stimulus? In these cases, crayfish usually display various forms of escape responses, such as repeated tail-flipping, walking rapidly away from the stimulus, or grabbing the stimulus with the non-touched claw (see Puri & Faulkes, 2015). This and other noxious stimuli also provoke physiological responses in crayfish, such an increase in blood glucose or changes in the level of serotonin (Fossat et al., 2014; Listerman et al., 2000; Puri & Faulkes, 2015). Moreover, when attacked or threatened either by another crayfish or any type of visual motion a crayfish detects, the animal assumes a defensive stance: orienting the body to face the stimulus, raising the thorax, and extending and opening the claws (Kelly & Chapple, 1990; Listerman et al., 2000). In contrast, there is limited evidence of protective behavior in crayfish (see Puri & Faulkes, n.d.; Thorp & Ammerman, 1978).

Crayfish also show various forms of learning. For instance, they are able to associate a light signal with a shock and learn to avoid the shock by walking forward to the other end of a shuttle box, rather than naturally tail-flipping (Kawai et al., 2004). They can even learn other unfamiliar actions, such as pulling a lever to obtain food (Olson & Strandberg, 1979 in Perry et al., 2013). In addition, crayfish may use different strategies—e.g. place cues or response learning—to solve spatial tasks. They can learn and remember the route out of a maze even one week after their initial training trials, suggesting that spatial memories in crayfish are relatively enduring (Tierney & Andrews, 2013).

Remarkably, crayfish appear to experience valenced emotional states. They exhibit stress-induced avoidance behavior strikingly similar to vertebrate anxiety. In an experiment, crayfish were exposed to an unpleasant electric field. The animals were then placed into a cross-shaped tank. Two of the arms of the cross were dark—an environment that most crayfish prefer—while two were lit. A crayfish that has not been exposed to the electric field will go to all of the arms, but with a slight preference for the dark areas. However, after being exposed to the electric field, the crayfish avoided the aversive, illuminated arms, since they were probably perceived as threatening. This behavior persisted in the long-term—for at least 24 hours. These animals had increased levels of serotonin, which is a known neurotransmitter released by the brain to counteract anxiety. In the same study, it was observed that after being injected an anxiolytic drug, the crayfish appear to calm down: only then they entered the dark arms of the tank as normal, and began to explore the lit arms as well. (Fossat et al., 2014).

Furthermore, crayfish also display stereotypic behavior when given methylenedioxymethamphetamine (MDMA) and other psychostimulant drugs (Huber et al., 2018; Panksepp & Huber, 2004). They are also negatively affected by social isolation, showing—among other effects—more agonistic interactions similar to those described in socially deprived vertebrates (Hemsworth et al., 2007; Patoka et al., 2019). In the absence of defeaters, these behaviors are evidence that crayfish probably experience negatively valenced emotional states.

Finally, opioid-like receptors have been identified in crayfish (see e.g. Nathaniel et al., 2010). However, the potential analgesic effects of opioid substances on these animals has not yet been studied. For their part, recreational drugs and anxiolytics have similar effects to those observed in humans, suggesting that some of the neural and chemical pathways which these drugs affect are well preserved in evolution (e.g. Fossat et al., 2014; Fossat et al., 2015; Huber et al., 2018; Jackson, & van Staaden, 2019; Panksepp & Huber, 2004).

Ants

In general, our knowledge about ants' capacity for consciousness is more restricted than for the species or taxa described above. We did not find scientific research for almost half of the studied features (48% of them). Hence, although most of the current evidence indicates positive responses for different potential consciousness indicators, the lack of research for other features suggests that our findings about ants (family Formicidae) possess greater uncertainty than our findings for other species or taxa.

Despite this limitation, almost all the reviewed literature suggests positive responses for ants—50% of the features qualify as “likely yes” or “lean yes”. In general terms, the existing knowledge highlights that ants, like honey bees and other social insects, display complex behaviors. These patterns seem to be correlated to their brain size—in a study with Atta colombica ants it was seen that the smallest ants had brains that constitute 15% of their body mass (Seid et al., 2010). Physical characteristics and functional descriptions of ants’ brain point out a centralized information processing (see e.g. Gronenberg, 2008; Hölldobler, B., & Wilson, 1990; Sømme, 2005). However, it is still unknown whether these insects have nociceptors or equivalent structures.

Are ants’ complex behaviors automatic expressions of pre-programmed responses? Or is it more plausible to suppose that they are programmed for general actions complemented by learning? Exploring the degree of automation and flexibility of ant behavior can shed some light on this matter. In this regard, we found that ants—similarly to honey bees—display contextual learning using specific and plastic cues (Perry et al., 2013). For instance, in an experiment, ants (Camponotus aethiops) learned to associate a certain cuticular hydrocarbon profile (the scent of an ant from another colony) with food reward and approached the food accordingly. However, when encountering a non-nestmate ant carrying the odour, the trained ants recognised the profile as indicating a “competitor” and became aggressive in order to defend their colony. This suggests that ants, like some, but not all other insects, “show interactions between different modalities (i.e. olfactory and visual), and can treat complex chemical cues differently, according to the context in which they are perceived” (Bos et al., 2010: 839).

Ants, as eusocial insects, need to exchange information and behave accordingly in order to meet the colony’s needs. In this regard, different studies have documented that ants lead other younger ants from the nest to food sources, or guide other ants to a new food source that the first ant would have found. This technique, called ‘tandem running’, is also used to find and choose better, new nest sites to which the colony can emigrate. Various authors state that this behavior is a form of social learning (Franks & Richardson, 2006; Stroeymeyt et al., 2017). However, other researchers claim that tandem running would prove that ants communicate with each other—rather than being a form of observational learning (Leadbeater et al., 2006). In any case, it seems that tandem running requires a certain ability to adapt to context—e.g. discriminating among potential information recipients—which would seem implausible in nonconscious individuals that respond to automatisms.

Furthermore, ants use a wide variety of materials in order to transport water or liquid food to their nest—including artificial materials they would have never encountered in nature, but were introduced as part of an experiment. They use different tools to transport different types of substances (for example, honey versus honey mixed with water), and they also learn to use certain materials preferentially (Maák et al., 2017). The use of debris as tools for transporting food has been documented among various species of ants.

Similarly to honey bees, there is partial evidence that ants could be aware of their certainty or uncertainty when making a choice (uncertainty monitoring). In an experiment with a T-maze, researchers trained ants to a feeder location, and then altered the environment by moving the feeder to the other arm of the T-maze. After finding the new food source, ants upregulated pheromone deposition as if they had made a wrong choice. Additionally, the researchers found that ants heading towards the food source which made an error deposited less pheromone. This seems to imply that the ants can measure the reliability of their own memories, and respond accordingly by depositing more or less pheromones. However, the authors are uncertain about what their findings represent. They write that "it's hard to believe that such tiny-brained animals are capable of such an advanced cognitive feat" and "one could conceive of several alternative explanations for our findings, which do not invoke metacognition" (Czaczkes & Heinze, 2015: 5). At the same time, they argue that their findings, “alongside similar results from honeybees (Perry & Barron, 2013) are suggestive of metacognitive abilities in social insects” (Czaczkes & Heinze, 2015: 5). Further research about uncertainty monitoring in ants and on their cognitive skills in general is needed.

By the same token, except for the significant adverse effects of social isolation (which seems to negatively impact ants’ health and results in aggressive interactions), there is no additional evidence on mood state behaviors in ants.

Finally, ants are considered “expert navigators”, and can exhibit considerable flexibility when integrating contextual cues (see e.g. Graham & Philippides, 2017; Wehner, 2003). In contrast, little is known about ants’ pain system and noxious stimuli reactions, their motivational tradeoffs and their responses to various drugs.

Cockroaches

Regarding cockroaches (genus Periplaneta), the existing literature is limited. No evidence was found (“unknown” responses) for an important part of the studied features—49% of them—which poses a greater degree of uncertainty in interpreting our results on these insects. We did not encounter any direct evidence against any of the features, but indirect research suggests that four features (8.1%) are not observed in cockroaches or that there is a low chance that it is found in them—tool use, flexible tool, play behavior and adverse effects of social isolation. We found positive evidence (“likely yes” and “lean yes”) for 42.9% of the reviewed features.

A cockroach has nearly 1,000,000 brain neurons (van Huis, 2017), more than a fruit fly, an ant and even a bit more than a honey bee. Cockroaches’ central nervous complex processes spatially structured mechanosensory and proprioceptive information (Guo & Ritzmann, 2013), which suggests that they centralize information processing. Additionally, cockroach brains, as vertebrate brains, apparently function in the control, organization, and planning of walking (Zill, 2010). However, parts of their nervous system that are not the anterior brain are also capable of controlling breathing, movement and learning (EFSA, 2005).

No studies indicating whether or not these insects possess nociceptors were found. However, given the presence of opioid-like receptors, various analgesics (morphine, promedol) affect cockroaches in a similar manner in which these substances affect humans. In particular, morphine, naloxone, and other analgesics, increase the time cockroaches can stay in a hot camera (Gritsai et al., 1998; Gritsai et al., 2004). Moreover, cockroaches exhibit various reactions to noxious stimuli: for example, a cockroach will respond to a tactile or wind stimulus directed at the body by turning away from the stimulus and running away. This escape response due to wind puffs, which may signal a predator attack, is a well-studied phenomenon (Bell, 1981; Domenici et al., 2009). If cockroaches are injected with parasitic wasp venom, they will groom, attempting to rid themselves of the fluid (Gal et al., 2005; Weisel-Eichler et al., 1999).

The most abundant evidence about cockroaches as potential conscious individuals are learning indicators. Cockroaches do not only show simple forms of learning—such as sensitization, habituation and classical conditioning—but they can also be conditioned to perform unfamiliar actions, such as moving one of their legs to a specific position in order to avoid a shock (Eisenstein & Carlson, 1994). Cockroaches can also learn to discriminate between different associations of a stimulus, depending on the context in which it is presented (contextual learning). In an experiment, cockroaches learned to associate one odor (vanilla) with water and another (peppermint) with saline under illumination, but in the dark to associate the opposite—vanilla with saline and peppermint with water (Sato et al., 2006).

Furthermore, cockroaches can learn to avoid noxious stimuli—like electric shocks—and remember it for at least 24 hours (Chen et al., 1970; Freckleton & Wahlsten, 1986). Gregarious cockroaches (of a different genus, i.e. Blattella germanica), for their part, learn social codes and how to interpret their conspecifics while interacting with others (Lihoreau et al., 2009).

Cockroaches display most of the studied navigational skills. For instance, in the laboratory, cockroaches learn mazes and can be rapidly trained in place learning tasks (Webb & Wystrach, 2016).

Additionally, cockroaches can show signs of learned helplessness. Following inescapable shock training, cockroaches often become passive and still when confronted with an escapable shock (Brown & Stroup, 1988).

Despite the presence of some anatomical features and noxious stimuli reactions in cockroaches, other numerous features in these insects merit further investigation.

C. elegans

Caenorhabditis elegans are small-size (1.5-mm-long when adult), free-living nematode worms that populate temperate soil environments. These nematodes have been widely used to study genetics and developmental biology (Riddle et al., 1997). Hence, it is not surprising that there is solid knowledge about their nervous system and other related features. However, existing research on C. elegans has focused on genetic analysis, and thus, no evidence was found for 44.9% of the features. Given C. elegans’ simplicity it is also probable that researchers may consider that several of our consciousness-indicating features are not likely to be found in these worms. On the other hand, we encountered positive evidence (“likely yes” and “lean yes”) for less than half (40.8%) of the reviewed features and some of them can occur unconsciously in humans^[5]. The other 14.3% of features were not observed or there were reduced chances (below 50%) that they are found in these animals.

C. elegans are one of the simplest organisms with a nervous system. In the hermaphrodite, this system comprises 302 neurons (White et al., 1986), the lowest number of nervous cells among all studied animal taxa^[6] and around 33 times less than the low end range of 10,000 neurons for sea hares. Their interneurons (neurons that transmit impulses between other neurons) integrate sensory information with ongoing network states to produce neuronal and behavioral variability (Gordus et al., 2015). The lateral ganglion and other neural circuits may provide an essential connection between the sensory and motor components of the C. elegans nervous system, which suggests a degree of information processing centralization (see Chatterjee & Sinha, 2008; Ghosh et al., 2016). Nevertheless, unlike other invertebrates, C. elegans do not seem to have a specialized neural region for the processing of spatial information and organization of movement. In other words, movement and stimuli discrimination do not appear to be integrated in a manner sufficiently similar to the vertebrate midbrain (see Altun & Hall, 2011; Kato et al., 2015).

Nevertheless, nociceptors have been precisely mapped in C. elegans (see e.g. Himmel et al., 2017; Smith et al., 2010; Sneddon, 2018). Congruently, some noxious stimuli reactions have been identified, specifically, physiological responses to nociception and moving away from a noxious stimulus. For instance, when touched, C. elegans escape the stimulus by turns and reversals or accelerated forward motion. When touched at the anterior body, C. elegans move backwards, and when touched at the posterior body, these worms move forward (Altun & Hall, 2011). Comparable reactions were found when C. elegans were exposed to heat (Leung et al., 2016; Mohammadi et al., 2013; Wittenburg & Baumeister, 1999). This heat-evoked escape response in C. elegans is considered highly stereotypical and a reflexive reaction (Leung et al., 2016). This response can be reversed with the administration of an analgesic substance (Nieto-Fernandez et al., 2009).

The presence of opioid-like receptors in C. elegans and hence, the effect of analgesic substances support the existence of an opioid-mediated modulatory mechanism for nociception in these worms and in invertebrates in general. Additionally, C. elegans are affected by recreational drugs in a similar manner to humans: in an experiment, it was observed that after being exposed to alcohol, C. elegans display disinhibited locomotion and feeding behaviors (Topper et al., 2014). For its part, the antianxiety drug sertraline stops C. elegans’ fear-like responses (turning around and actively moving away), induced by the presence of a predator (Liu et al., 2018). However, antidepressants does not appear to provoke similar responses to those caused by these substances in humans (Ford & Fong, 2015). Broadly, the above findings suggest that some of the neural and chemical pathways that these drugs affect are preserved in evolution (Davies et al., 2015; Liu et al., 2018; Topper et al., 2014).

For their part, some learning indicators are exhibited by C. elegans. The simplest forms of learning—i.e. classical conditioning, sensitization and habituation—have been observed in these worms (see e.g. Hedgecock & Russell, 2007; Kimura et al., 2010; Morrison et al., 1999; Wen et al., 1997). Surprisingly, C. elegans can probably learn and remember whether a food source was pathogenic (infected with a pathogenic bacteria), avoiding the infected food. This conditioned avoidance reaction can be considered a form of taste aversion behavior (Reddy et al., 2009; Zhang et al., 2005). In another experiment, it was established that C. elegans also show long-term memory for the “tap-withdrawal response”, a behavior alteration to avoid a potential noxious stimulus. This response was observed even 24 hours after the nematode had been trained (Beck & Rankin, 1997; Rose et al., 2002).

There is no significant evidence for motivational tradeoffs, sophisticated cognitive skills and mood state behaviors in C. elegans, except for the adverse effects that social isolation provokes in the worm’s development and behavior (Ardiel & Rankin, 2009; Rai & Rankin, 2007; Rose et al., 2005).

Finally, despite being a widely studied animal, it should be noted that evidence of consciousness in C. elegans is much weaker than in fruit flies, another animal commonly employed in biology research.

Sea hares

Sea hares (genus Aplysia) are some of the most active marine gastropod molluscs. Much research has been carried out on the nervous system of Aplysia. However, the existing literature on sea hares as conscious animals is limited. No evidence was found (“unknown” responses) for almost half of the studied features—48% of them—which poses an important degree of uncertainty in interpreting our results on these molluscs. We did not encounter any direct negative evidence against any of the features, but other research suggests that 12% of the features are not observed or that they have low chances to be found in these animals. Positive evidence (“likely yes” and “lean yes”) was identified for 40% of the reviewed features.

Aplysia’s central nervous system is composed of nine ganglia. The abdominal ganglion of Aplysia californica has a relevant role in the relatively centralized processing of information (Frazier et al., 1967). However, it is not known whether they have a specialized neural region for the processing of spatial information and the organization of movement.

Primary nociceptors have been investigated extensively in Aplysia (Walters & Moroz, 2009; see also Illich & Walters, 1997). Outstandingly, interesting similarities exist in the behavioral responses to noxious stimulation of Aplysia and mammals. In particular, Aplysia displays the nearly ubiquitous pattern of immediate withdrawal reflexes, rapid escape, and prolonged recuperative behaviors exhibited across all major phyla (Crook & Walters, 2011). Noxious stimuli—such as elevated temperatures and other environmental stressors—probably induce stress proteins and other physiological responses in Aplysia (Carefoot, 1994; Greenberg & Lasek, 1985; Sanders, 1993). Moreover, when disturbed, sea hares release a noxious ink, which qualifies as a defensive behavior (Aggio & Derby, 2010; Kicklighter et al., 2007).

In addition, sensitization and habituation of the withdrawal reflex have been demonstrated in sea hares (Carew et al., 1971; Carew & Sahley, 1986; Pinsker, et al., 1973). Sea hares can learn to associate a chemosensory stimulus (a shrimp extract) with a head shock. Later, the shrimp extract greatly facilitates withdrawal responses and escaping (Walters et al., 1979 in Carew & Sahley, 1986). In an operant conditioning experiment, Aplysia learned to bite in order to receive electrophysiological stimulation to the anterior branch of the esophageal nerve (which presumably conveys information about the presence of food during ingestive behavior) (Brembs, 2003). Sea hares can also learn that certain environments are more dangerous than others and will more readily initiate defensive and withdrawal behaviors in them for at least 48 hours following conditioning (Colwill et al., 1988).

Noxious stimuli can also cause the animal to suspend feeding (Walters et al., 1981). This reaction to electric shocks, along with other responses such as head withdrawal, inking and escape locomotion, resemble states in mammals that have been defined as conditioned fear. Hence, it could be argued that this induced state in sea hares satisfies the function of fear as a general, preparatory defensive state elicited by stimuli signaling imminent danger. Similarly, feeding behavior is also inhibited when sea hares are reared in isolation from their conspecifics (Schwarz et al., 1998; Schwarz & Susswein, 1992).

In addition, studies on Aplysia have substantiated the claim that opioid receptors are found in molluscs and may show a degree of receptor localization not yet demonstrated in mammals. (Harrison et al., 1994; Leung et al., 1986). Consistently, an analgesic substance (FMRFamide) reduces excitation and synaptic transmission in sea hares, resulting in decreased withdrawal reflexes (Small et al., 1992 in Sneddon, 2015).

Although much research has been carried out on the nervous system of sea hares, little is known about the neuroanatomical loci and specific neurons involved in some of their more complex responses—e.g. those related to long-lasting learning and to a fearlike state. Additionally, some researchers also point out that sea hares’ responses would be too rigid to be considered conscious (EFSA, 2005).

Spiders

As previously pointed out, we investigated spiders at the rank of order (order Araneae), which contains approximately 45,000 species (World Spider Catalog, 2019). Compared with the other considered taxa, this is the invertebrate category that by far comprises the largest number of species. Given the above, any generalization for the whole order is provisional and should be considered with caution.

In general, most of the existing literature on consciousness-indicating features in spiders is relatively reduced and recent. Thus, no evidence was found (“unknown” responses) for more than half of the revised features (57.1%). We did not encounter any direct study for a negative response for any of the features, but indirect research suggests that three features (6.1%) are not observed or that they have low chances to be found in these animals. We found positive evidence (“likely yes” and “lean yes”) only for a bit over a third of our indicators (36.7% of the features).

Spiders, along with cephalopods, crustaceans and some insects, have notably complex brains (see Foelix, 2011). Their central nervous system has some resemblance to those of crustaceans and insects, but is even more concentrated, suggesting centralized information processing (Brusca & Brusca, 2002 in Sømme, 2005). In spiders, the ventral ganglionic mass and its related nerves can be identified as structures specialized in the processing of spatial information and organization of movement (Barth, 2002).

It is not yet known whether these arachnids have nociceptors or not. However, spiders do react when disturbed or attacked, by moving away, jumping or dropping to the ground and displaying 'wheeling locomotion', depending on the species. They also defend themselves from predators, and if they are stung in a leg by venomous insect prey, spiders will shed (autotomize^[7]) that leg (Cloudsley-Thompson, 1995; Foelix, 2011).

Habituation, sensitization and associative learning have been demonstrated in spiders. For example, they can associate different stimuli, such as prey, with color cues. In an experiment, Phidippus princeps spiders learned the location of prey in a T-maze by color cues alone (Jakob & Skow, 2007). In another study, spiders were confined in an island surrounded by water, and they had to choose between two potential escape tactics (leap or swim), one of which would fail. Using trial-and-error, spiders could escape from the island both when correct choices were rewarded and when incorrect choices were punished (Jackson et al., 2001). These two experiments are not only examples of conditioning in spiders, but they also point out spiders’ navigational skills, including detour—a behavior that certainly requires flexibility. Moreover, a recent investigation has established that spiders can remember and avoid a noxious stimulus, even 24 hours after being trained (Peckmezian & Taylor, 2017). In sum, learning skills, spatial memory, along with path integration and navigational abilities, suggest that spiders are not entirely instinctual and they do display behavioral plasticity (see Jakob et al., 2011).

Despite the fact that most spider species are asocial, spiders seem to engage in observational learning as well. In a study, it was seen that they would copy their conspecifics’ food preferences to some extent (Adams, 2015). Likewise, growing up in contact with conspecifics may prompt spiders’ cognitive development and exploratory behavior (Liedtke et al., 2015; Liedtke & Schneider, 2017). Nonetheless, we did not find further evidence about spiders’ cognitive skills and mood state behaviors. For instance, evidence for tool use among spiders is very limited (see Henschel, 1995).

Despite being barely studied, in general, available behavioral studies suggest that spiders are possibly conscious. Yet, our knowledge of their pain system (assuming they have any) is still limited. Further research in spiders is needed.

Earthworms

Earthworms (Lumbricus terrestris) have much simpler nervous systems than cephalopods, crustaceans, arachnids and insects. In this case, we face as well a significant degree of uncertainty—there is no available evidence for more than half of the features that we considered in our study (52%). We found positive evidence (“likely yes” and “lean yes”) only for a bit over a third of our indicators (36% of the features). We did not encounter any direct study with negative evidence against any of the features, but indirect research suggests that six features (12%) are not observed or that they have low chances to be found in these animals.

The whole nervous system in these annelids seems to present a very rudimentary condition. However, earthworms do possess a relatively centralized nervous system (Lewis, 1898; Sømme, 2005) and it has been discovered that the cephalic ganglia has a role in the processing of internal and external stimuli (McManus et al., 1982). However, given the segmentation of the nervous system, different ganglia control local portions of the body of the worm and these segments are not directly connected to each other (see Prosser, 1934). Thus, specific ganglia respond to sensory information from the local regions of the body and control the local muscles (Chakravarthy, 2018), which entails a low centralization in the processing of spatial information and organization of movement.

It is not known yet whether earthworms possess nociceptors or not. Yet, some noxious stimuli reactions have been identified in these annelids. For instance, earthworms secrete a mucus that acts as a lubricant and—as it is rich in nitrogen—it functions as a protective coat to noxious stimuli as well (Edwards & Bohlen, 1996). It has been noticed that earthworms secrete mucus in large amounts when the animals are immersed in a noxious stimulus such as acid (Laverack, 1963). Furthermore, other noxious stimulation like handling, pinching, stimulating earthworms with electric shock and severing them elicits copious secretion of mucus (Ressler et al., 1968). In addition, before a threatening stimulus, escape or startle reflexes are common reactions observed in earthworms (Drewes, 1994).

Other physiological capacities have also been identified in earthworms. In particular, opioid-like receptors have been found in these animals (Alumets et al., 1979; Renzelli-Cain & Kaloustian, 1995) and injections of naloxone (an opioid antagonist) inhibit the touch-induced withdrawal reflex (Gesser & Larsson, 1986). These findings suggest that opioid substances play a role in earthworms’ sensory modulation, similar to that found in many vertebrates (see Smith, 1991).

Furthermore, it has been demonstrated that starvation alters risk-taking behavior in earthworms, a predator avoidance tradeoff. First, it must be considered that these animals are are negatively phototactic and thus, when feeding, they will prefer dark environments. Consistently, high-light conditions suppose a risky situation, since these conditions expose worms to desiccation and to be easily spotted by predators. However, when starving, earthworms will make riskier choices, opting for a high-food and light spot, instead of a safer (dark) but less nutritious condition. These results indicate that earthworms make state-dependent foraging choices (Sandhu et al., 2018).

Earthworms are also able to display simple forms of learning, such as sensitization (Peeke et al., 1967), habituation (Johnson, 1970; Kuenzer, 1958; Ratner & Gilpin, 1974) and classical conditioning (McManus & Wyers, 1979; Ratner & Miller, 1959). In an instrumental conditioning paradigm, earthworms learned to distinguish between the limbs of a T-maze to gain access to moist earth or moss, rather than a sandpaper floor and an electric shock. (Arbit, 1957 in Gregory, 2008; see also Jacobson, 1963). In a similar experiment, it was observed that the learned path and avoidance of the injurious stimulus persisted over a three-week period of no practice and after removal of the esophageal ganglia (Yerkes, 1912 in Datta, 1962). However, it must be noted that, in general, earthworms’ responses to environmental stimuli are highly stereotyped and inflexible. This fact has led to some disagreement over the interpretation of the studies cited above. Regardless, earthworms’ abilities to learn seem to be restricted.

Given the current state of our knowledge, there is a lot of uncertainty about whether or not earthworms are conscious. First, their pain system (if they have any) has not been sufficiently studied. Second, in spite of certain behavioral evidence, it is hypothesized that earthworms’ reactions to noxious conditions may be reflexes (see e.g. Sømme, 2005).

Moon jellyfish

Of all the species or taxa studied, the highest degree of uncertainty is found for moon jellyfish (genus Aurelia): in this case, the "unknown" answers amount to 64% of features. Regarding the remaining features—those for which there is direct or indirect evidence—unlike previous categories, most of the responses are negative. More specifically, the literature suggests that 12 features (24%) have not been observed or that they have low chances to be found in these animals, a number that duplicates the proportion of positive evidence (“likely yes” and “lean yes” responses), which, in turn, accounts for 12% of the reviewed features. The proportion of negative evidence and the lack of research on most of the features suggest that the chances of jellyfish being conscious are low.

This hypothesis is congruent with the fact that jellyfish do not appear to have a centralized nervous system. Instead, jellyfish possess a very simple radial network of cells known as a nerve net. This net spreads around the inner margin of the bell and interacts with clusters of small sensory structures (rhopalia). Although there is some disagreement, it is generally held that this nervous net has a diffuse organization (see Albert, 2011; Johnson et al., 1994; Katsuki & Greenspan, 2013; Mackie, 2002; Satterlie, 2014; Petie et al., 2011). Consistently, there is no centralization in the processing of spatial information and organization of movement in jellyfish (Mackie & Meech, 1995a, 1995b, 1995c; Satterlie, 2002).

Jellyfish do possess various sensory receptors, but there is no published evidence supporting the existence of strict nociceptors (Albert, 2011). However, they exhibit some noxious stimuli reactions. For instance, stationary Aurelia sp. jellyfish swam rapidly away and increased their pulse rate when touched by a predator (Hansson, 1997; Hansson & Kultima, 1996; Strand & Hamner, 1988). They also release a paralyzing toxin for prey capturing and for protecting from potential predators (sting response) (Katsuki & Greenspan, 2013; Lecointre & Le Guyader, 2007).

Except for habituation (Johnson et al., 1994), there is no evidence of other learning indicators in jellyfish. The same applies to motivational tradeoffs, cognitive sophistication features, and drug responses. There is restricted evidence about jellyfish’s navigational skills. Existing literature suggests that there are low chances that Aurelia jellyfish direct their navigation towards a target place or for a specific purpose (Albert, 2011; but see Katsuki & Greenspan, 2013).

Reduced possibilities of centralized information processing, no current evidence about jellyfish’s pain system (if they have any), and their behavioral simplicity, suggest that there is a low probability that jellyfish are conscious.

Protists, prokaryotes and plants

As explained in a previous post, we investigated plants (kingdom Plantae), prokaryotes and protists to give a sense for how often the studied features are found in organisms widely believed to lack consciousness. Indeed, protists, prokaryotes, and to a greater extent, plants are the categories for which we found higher percentages of evidence of negative responses (“likely no”). These findings are presented below.

Protists are a very diverse group of organisms. Historically, this category groups eukaryotic organisms (those with cells containing a nucleus) that do not fit the criteria for the kingdoms Animalia, Fungi, or Plantae. Besides their relatively simple levels of organization, protists do not necessarily have much in common: they live in a variety of aquatic and terrestrial environments, and occupy many different niches. Many are single-celled, although some phyla of protists include descendants that are multicellular or colonial (single cells that live together as a unit) (Alters, 2000). Given the current state of the art and, because of its relevance, most of our research focuses on protozoans (animal-like protists^[8]). In particular, we have focused on paramecia, a genus of unicellular ciliates.

We observe that no evidence was found (“unknown” responses) for a considerable percentage of the studied features (42.9% of them). That is probably because these indicators are very unlikely to be found in these organisms, given the widespread belief that these organisms lack consciousness. Consistent with this hypothesis, the current literature suggests that a significant part of the features are not observed or there are limited chances (below 50%) that they are found in protists^[9].

In this regard, an important objection against the presence of consciousness in protists is the fact that these organisms have no neurons and therefore, no nociceptors. However, there is conflicting evidence regarding the presence of nociceptive behavior in protists. Commonly, paramecia move away from noxious stimuli (Boisseau et al., 2016; Jennings, 1906; Wood, 1969; Zupanc, 2010). Therefore, it can be argued that they satisfy a loose definition of nociceptive-type responses (e.g. Smith, 1991). According to Kavaliers (1988), these responses appear to be elicited by mechanisms that are reminiscent of nociceptors. However, phobotaxis is a random behavioral response that does not clearly integrate the direction of the stimulus.

Besides moving away from noxious stimuli, chemical changes may be also induced in protists when subjected to environmental stressors (Slaveykova et al., 2016). Furthermore, Tetrahymena has opioid-like receptors (see e.g. Harrison et al., 1994). Protists are motile, navigate known and unknown areas, and some of them (the slime mold) seem to use an externalized spatial memory for simple bounded environments (Reid et al., 2012; Smith-Ferguson et al., 2017). Moreover, they show simple, and usually considered automatic, forms of learning—sensitization, habituation and, probably, classical conditioning.

All the features for which we found positive responses for protists (“likely yes” and “lean yes”) sum up to 24.5% of all the studied indicators. Nevertheless, almost all these positive responses show no plasticity and appear to be automatic reactions. Additionally, the most sophisticated features found in protists (i.e. physiological responses to nociception or handling, movement away from noxious stimuli, sensitization, habituation and classical conditioning) occur frequently nonconsciously in humans (see our upcoming report). Hence, displaying those features is insufficient grounds for consciousness in protists.

Regarding prokaryotes, we found even weaker evidence that they display the studied features. Prokaryotes are unicellular organisms that lack a membrane-bound nucleus, mitochondria, or any other membrane-bound organelle. It includes two domains of organisms, Archaeabacteria and Eubacteria (Alters, 2000.

Similarly to what was described for protists, here we observe that no evidence was found (“unknown” responses) for a considerable part of the studied features (42.9% of them). This is probably because these indicators are very unlikely to be found in organisms that are assumed to lack conscious awareness. Consistent with the above, the current literature suggests that most of the features are not observed or there are reduced chances (below 50%) that they are found in prokaryotes^[10].

Unlike invertebrates, these simple organisms have no neurons and no nociceptors. However, Escherichia coli bacteria possess mechanosensitive channels that allow the cell to prevent its destruction (Levina et al. 1999; Sukharev et al. 1994). Furthermore, through chemotaxis, bacteria move from different chemical gradients, helping the organism to find optimum conditions for its growth and survival (Pandey & Jain, 2002). Although these mechanisms may qualify as nociceptive reactions, it must be considered that they are automatic responses, merely elicited by environmental chemical changes.

Aside from the above, sensitization and probably habituation have been documented in bacteria (Koshland et al., 1982; Lyon, 2015; Porter et al., 2011; Stock, 1999).

In sum, all these positive responses for prokaryotes (“likely yes” and “lean yes”) amount to 10.2% of all the studied indicators. The above points out, as it is known, that bacteria are able to respond to their environment in order to survive. However, there is not enough evidence to justify the claim that these organisms might be conscious.

Regarding plants, these are mainly multicellular, predominantly photosynthetic, eukaryotes of the kingdom Plantae. As it happens with protists and prokaryotes, no evidence was found (“unknown” responses) for most of the studied features (43.8% of them). As before, this is probably because these indicators are very unlikely to be found in these organisms. Since there is no evidence for structures such as neurons, synapses, a brain or equivalent organizations in plants (see Alpi et al., 2007), the literature directly or indirectly states that most of the features are not observed or there are reduced chances (below 50%) that they are found in plants^[11].

Nevertheless, it is widely known that plants have the ability to sense, and respond to, their environment to adjust their morphology, physiology, and phenotype accordingly. Plants, in fact, react to some potential noxious stimuli through physiological responses (see e.g. Fincher & Stone, 1981), and in Mimosa pudica, for example, with rapid folding of leaflets (Amador-Vargas et al., 2014). These responses may be viewed as lax nociceptive reactions. However, the above does not imply per se that these are conscious responses, since there are no nociceptive structures in plants.

There is partial evidence that plants may be able learn. According to some researchers, these organisms show sensitization (Calvo et al., 2017), and probably, habituation (e.g. Gagliano et al., 2014) and classical conditioning (e.g. Gagliano et al., 2018). However, regarding habituation, Moore (2004 in van Duijn, 2017), among others, warns that this phenomenon, from protists and plants to vertebrates, is unlikely to be homologous. Something similar can be hypothesized with respect to sensitization and classical conditioning. In fact, it is also argued that the above findings account for plants’ adaptability, but are not necessarily evidence that plants "learn" (see Alpi et al., 2007).

In general, the features where we found positive responses for plants (“likely yes” and “lean yes”) are related to plants’ adaptability and ability to respond to their environment. As a whole, they sum up to 10.5% of all the studied indicators, comparable to the prokaryotes’ score –which are widely considered to lack conscious awareness. Despite some academics’ interpretations of current findings, it must be considered that the scientific community overwhelmingly reject plant consciousness. It is assumed that plant cells do share features in common with neurons, and signal transduction and transmission over distance is a property found in animals and in plants. However, no comparable structures for signal propagation at the cellular, tissue and organ levels exist in these organisms (see Alpi et al., 2007).

Conclusions

• In general, we can affirm that there are more unknowns than certainties about the examined indicators of consciousness in invertebrates. Of the 53 potentially consciousness-indicating features reviewed, we found no evidence regarding 43.6% of the cases. On the other hand, there is relatively strong evidence, either positive or negative, about around 35.5% of the features.
• Broadly, the biggest gaps of knowledge are related to cognitive sophistication, mood state behaviors, and motivational tradeoffs in invertebrates. Likewise, current studies about to what extent various drugs cause in invertebrates responses similar to those which they produce in humans are insufficient. Further research is needed about these issues.
• It must be noted that our findings may be limited by the preference for more “popular” species by researchers (‘taxonomic bias’), which results in the underrepresentation in publications of, and lack of knowledge about less popular animals like invertebrates (see Wilson et al., 2007). Moreover and as mentioned in a previous post, the tendency to underreport negative results (‘positive publication bias’, see Mlinarić et al., 2017) may discourage further research that shows invertebrates lack certain features they are assumed to lack. In this sense, we must emphasize that this paucity of evidence does not necessarily entail that invertebrates are “not conscious” or “less conscious” than other animals. Nor that they lack, for example, certain cognitive abilities or emotional states. Limited evidence of consciousness should not be confused with limited consciousness.
• Evidence supports the existence of nociception in invertebrates, but this conclusion is mostly based on behavioral observations (e.g. moving away from noxious stimuli), rather than in the identification of nociceptors connected to higher brain centers or equivalent structures. Indeed, nociception (loosely understood) is found across all the invertebrate taxa. However and with some exceptions (fruit flies, C. elegans, sea hares and octopuses), to date, it is not known whether or not different invertebrates have these structures, and if so, whether their functioning enables the centralised processing of information. As said before, absence of evidence is not equivalent to evidence of absence. This topic merits further investigation.
• Some pain system may be present in invertebrates. Behavioral and physiological responses to handling and to noxious stimuli, effects of analgesics, and the presence of opioid-like receptors in various taxa suggest that the fundamental physiology and biology of nociception are conserved in invertebrates. Additionally, we found important evidence that–to varying degrees–most invertebrate taxa may centralize sensory information processing. Hence, the above indicates that further research in this regard might be fertile.
• Behavioral evidence demonstrates that invertebrates may process noxious stimulation intensity and direction. Reactions such as escaping, fighting back and avoidance outline that they may not be simple automatisms. This idea is also supported by the ability of various invertebrate taxa to navigate their surroundings in a flexible manner.
• Nevertheless, most invertebrates do not exhibit behavioral responses in conditions in which we would expect great responsiveness from vertebrates, like vocalizations and protective behaviors towards injured body parts, such as wound rubbing or limping after an injury.
• It must be taken into account that some authors claim that invertebrate behavior–especially, molluscs (e.g. Aplysia) and in some cases, insects responses–would largely obey to pre-programmed patterns. This seems to be the case of various reactions found in earthworms and in C. elegans, for example. In insects, grooming behavior when they are in contact with a nocive substance appears to be an automatism.
• Invertebrates do learn. In general, they show simple forms of learning such as habituation, sensitization, and others based on associative mechanisms. Some invertebrates show more complex expressions of learning, such as operant conditioning with unfamiliar actions and social learning. That appears to be the clear case of honey bees, ants, crabs, crayfish and octopuses.
• Where’s stronger evidence? First, regarding the majority of invertebrates, it must be admitted that it is not possible, given the evidence presently available, to conclusively determine whether they are conscious or not. Nevertheless, of all the taxa we studied, current research has provided us with relatively strong evidence that cephalopods (i.e., octopuses) and crustaceans (i.e., crabs and crayfish) are conscious. In addition, evidence suggests that, probably, fruit flies are conscious as well. Furthermore, there is important behavioral evidence for the consciousness of honey bees and, to a lesser extent, ants, cockroaches and spiders. Their case, however, is not as strongly supported by existing evidence as that of cephalopods and decapods.
• Further conclusions about the precise likelihood of sentience and moral value of invertebrates are beyond the discussion we provide here, due to further considerations of sentience and moral weight. Ultimately further research will be needed, we will discuss these in more detail in future reports that aim to outline the current state of improving invertebrate welfare as a cause area.

Credits

This essay is a project of Rethink Priorities.

It was written by Daniela R. Waldhorn with contributions from Jason Schukraft, Peter Hurford, and Marcus A. Davis. Jason Schukraft, Peter Hurford, Marcus A. Davis, and Eze Paez provided helpful comments on this essay.

If you like our work, please consider subscribing to our newsletter. You can see all our work to date here.