Ascribing sentience to fish: potential policy implications

3. Evidence in relation to fish sentience

Most research has been conducted in relation to adult teleosts. We were able to find limited evidence specific to the wide range of fish species and to embryonic and early juvenile (e.g., before first feeding) forms.

1. Phylogenetic ‘proximity’ to other species who have been deemed sentient

Fishes are a paraphyletic group, with variation between species of fish across the three major groups: Agnatha (hagfishes, lampreys), Chondrichthyes (sharks, rays, sturgeons) and Actinopterygii (bony fishes). Of the last, most commercial aquaculture fish and those caught at sea for food are teleosts^[11]. Nonetheless, all fishes are vertebrates (with the taxonomic exception of hagfishes). As such they are within a phylogenetic clade of which all (healthy adult) members are deemed sentient (and which also includes humans), together with the most commonly kept farmed and pet species, and many animals used in science.

2. Neuroanatomical functioning

2.1 Evidence of central nervous system anatomy

The fish brain has a similar anatomy to that of other vertebrates, comprising a forebrain (telencephalon and diencephalon), midbrain (mesencephalon) and hindbrain (rhombencephalon). The fish brain is less complex in structure and smaller relative to body size than in mammals and does not appear to have a well-developed cerebral cortex^[12]. While this area plays a role in modulating emotions in mammalian brains, studies suggest that other parts of the fish brain may be responsible for generating emotional states^[13]. Further, sensory processing has been shown to occur in different regions of the brain depending on the fish species^[14,15].

Within the forebrain (telencephalon), fish lack the cortical centres seen in mammals that allow the conscious perception of pain. However, there are extensive connections between fore- and midbrains in fish that are associated in mammals with the perception of threats and injuries and the experiences of fear and pain^[16]. Other structures within the telencephalon associated with fear and pain perception in mammals are known to exist, particularly in teleost fish, and respond to noxious stimuli in a similar fashion^[17].

Within the teleost telencephalon, there are structures that appear to function similarly to the mammalian amygdala (important in emotions and arousal) and hippocampus (important for memory, learning and spatial relationships)^[18,19]. Fish also have a hypothalamus, which is known in mammals to be involved in sexual and social behaviours as well as acting as a junction and integration centre for signals from the telencephalon associated with fear responses^[20,21,22].

2.2. Evidence of complex sensory systems connected to the brain

Strong evidence has been available for some time that fish are able to perceive a number of stimuli, at a local and higher (telencephalon) level^[23]. These include very similar systems that are seen in other vertebrates, such as mammals and birds for example: good visual acuity (albeit adapted for an aquatic existence); auditory and vibration perception (through both ears and lateral lines); chemical communication through olfactory epithelium and taste buds (the latter located not only in the mouth but on the whole body surface); electrical and magnetic activity perception (through lateral lines and other mechanisms such as through specific cells in the basal lamina in the olfactory lamellae of rainbow trout^[24]). Some of these functions are ones that many other vertebrates have lost (e.g., magnetic and electrical pulse perception) and some are enhanced in comparison (such as having taste buds on the body surface). Therefore, fish appear as, and perhaps more, likely to be affected by some forms of environmental stimulation (particularly if adverse) than some mammal species.

2.3. Evidence of anatomy and physiology suggesting an ability to perceive pain

It has been shown that fish possess the anatomical components required to experience pain, such as functional nociceptors, and the required brain areas to allow processing of nociceptor stimulation, such as the pons, medulla and thalamus^[25,26,27]. Tracts of nerves, such as the trigeminal and spinothalamic tracts, are present in all groups of fish, similar to mammals, communicating noxious stimuli to the brain areas involved. Higher regions of the brain have been shown to have increased electrical activity specifically when noxious stimulation occurs, suggesting pain perception^[17].

Further support for pain perception exists in fish due to the presence of opioid receptors (similar to mammals) and endogenous opioid production, features widely used to prove whether a pain response can occur in an animal^{[28,29,30,31]}. Furthermore, administration of opioid agonists, that effect pain relief in mammals and birds, have been shown to be effective in reducing the severity of response to noxious stimuli in fish and the subsequent administration of opioid antagonists (e.g., naloxone) ablate this pain relief^[32].

It is also interesting to note that not only are endogenous opioids present in fish, but oxytocin or its analogue isotocin are also found in fish^[33]. Oxytocin in mammals is a hormone whose release is associated with pleasurable experiences and affiliative social interactions.

3. Behavioural indicators

3.1. Evidence of adaptive behaviours to aversive stimuli

The demonstration of an aversive response to a noxious stimulus supports the anatomical and chemical pathways for nociceptive function. In addition, adverse effects on the behaviour of an animal that go beyond reflex responses provide additional evidence of a psychological aspect suggestive of suffering and full pain perception by the animal. Fish may continue to demonstrate abnormal behaviours for hours after administration of a noxious stimulus^[34]. This includes avoidance of noxious stimuli, suggesting a learnt behavioural response – something which has been demonstrated in a wide range of fish species^{[35,36,37,38]}.

Fish demonstrate a fear response that suggests learning. It is known that learning in some other vertebrates is mediated in the brain with receptors activated by N-methyl-D-aspartic acid (NMDA), as demonstrated through the use of antagonists to NMDA that impair associative learning and conditioned fear. Injections of NMDA-receptor antagonists in goldfish (Carassius auratus) have been shown to block some of the aspects of Pavlovian fear conditioning in fish^[39].

Physiological responses associated with stress-inducing stimuli have been assessed in fish and found to be directly comparable to those seen in mammals and birds^{[40,41,42,43]}. Fish demonstrate coping strategies to mitigate both acute and chronic stress; some are proactive fight, fright and flight responses (based on adrenaline) and others passive, so-called ‘freeze and hide’ responses (largely based on cortisol)^[44].

3.2. Demonstration of cognitive, emotional and social behaviours

A number of scientific papers have demonstrated complex social behaviours which are frequently taken to relate to more complex cognitive abilities in mammals and birds. Clownfish (Amphiprion ocellaris) can recognise their own species (depending on stripe count) and demonstrate more aggressive behaviours to fish with the same number of white stripes as their own^[45]. In a related species of clownfish (A. percula), non-breeding individuals within a territory ‘queued’ behind the dominant breeding pair for the opportunity to ‘inherit’ the territory and to breed. These non-breeding queuing individuals showed neither aggression nor avoidance of others but waited patiently, demonstrating in the author’s opinion that they were prepared to tolerate non-breeding positions in their social hierarchy solely because of their potential to realise benefits in the future, i.e. the eventual gain of a breeding position^[46]. This may suggest a form of goal-directed decision-making and/or conceptualisation of time involved in decisions made now to yield benefits in the future.

Cichlids, such as angelfish (Pterophyllum sp.), show social behaviours including intraspecific aggression with territorial defence and strict hierarchies^[47,48]. These are established both in breeding and non-breeding individuals, and cues for social interaction are complex including visual and olfactory cues^[49]. When cichlids (e.g., Cichlasoma citrinellum) are placed in solitary isolation, the lack of social interaction leads them to become more immobile, lose their colour pattern and have increased fright responses^[50,51], indicating a negative mood state. The social context under which cichlids (Oreochromis mossambicus) are exposed to a novel object alters their behavioural response in a flexible way^[5], suggesting behavioural and motivational trade-offs. Cognitive performance in a discrimination learning test (T-maze performance) was also impaired with social isolation in the cichlid, Cichlasoma paranaense ^[53].

Fish have been shown to exhibit a range of complex social behaviours, such as cooperative behaviour^{[54,55,56,57]} and direct reciprocity^[58,59], social learning^{[60-66 ]} and preferences for kin^{[67 ]}or specific individuals^[68,69,70]. Female convict cichlids (Amatitlania siquia), which form monogamous pairs, show a negative “pessimistic” judgement bias (a bias that is usually interpreted as a low mood state), when paired with a non-preferred partner, in contrast to females assigned their preferred partner^[71]. In addition, there is evidence that social interactions may also provide a degree of buffering against stress in zebrafish (Danio rerio)^[72,73].

Some species of fish, such as Angelfish (Pterophyllum scalare), spend significant amounts of their time and energy in rearing their young, comparable to many mammals^[49]. This includes protecting the eggs from predators, fanning them to increase oxygenation, removing dirt, fungus and snails from the eggs, collecting eggs that fall away from the brood site and physically detaching the fry from the substrate that they are often stuck to by mucus attachments^[74,75].

Fish show significant non-social learning and capabilities. For example, they display spatial learning about their locations (e.g., for navigation)^[76] and familiarity of their environment^[77]. Guppies (Poecilia reticulata) can solve complex mazes, using similar responses to those seen in rats^[78]. Tool use has been observed in wrasse (Choerodon schoenleinii) (use of anvils)^[79] and archer fish show complex use of water as a tool to shoot down prey. Visual discrimination has been demonstrated in many fish species and cognitive styles have been shown in archer fish (Toxotes chatareus) (e.g., fast fish learn to shoot novel targets more quickly, but slower fish have better discrimination^[80]). Fish have also been shown to use referential gestures – coral reef fish signal presence of hidden prey to cooperative hunting partners^[81]. This is as part of rare interspecific group hunting, shown in fish but rarely observed in mammals or birds, where groupers (Plectropomus spp.) and moray eels (Gymnothorax javanicus) show signalling and intentional communication when hunting and both partners benefit from the association^[82].

Angelfish demonstrate non-symbolic numeracy, a feature identified in primates. A study^[83] subjected humans and the angelfish (Pterophyllum scalare), to the same numeracy tasks evaluating both absolute values (i.e. specific quantity for a reward) and relative numerosity (i.e. associations linked to abstract concepts such as bigger or smaller). Both angelfish and humans preferred relative numerosity when making choices and both learnt more rapidly to preferentially select ‘more’ rather than ‘less’ suggesting an adaptive response. The authors concluded that this relative numerosity was commonly seen in a wide range of vertebrates, including fish, and that all vertebrate species may share a system for making decisions about quantities.

Furthermore, there is behavioural evidence of learning by association with harms^{[21, 84-87]}, which can be modulated by anxiolytics^[88,89,90]. Goldfish exhibit complex motivational trade-offs between feeding and electric shock, adjusting their behaviour depending on shock intensity and level of feeding deprivation^[91]. In mammals and birds, this is interpreted as a cognitive response and is evidence of behavioural complexity beyond simple reflex behaviour.

4. Qualitative Behavioural Assessment (QBA)

QBA is a ‘whole animal’ approach to assess animal emotional expressivity, based on describing and quantifying the dynamic style of behaving, using descriptors such as relaxed, agitated, lethargic and confident. Use of this approach with farmed salmon has demonstrated that fish farmers and other observers were able to describe different dimensions of emotional response in juvenile farmed salmon with good agreement between them^[92]. Further, when salmon were exposed to a stressor, observers blinded to treatment were able to detect changes in emotional expression, with salmon perceived as more ‘unsettled/stressed/skittish’ and less ‘relaxed/content’ after sampling than before^[93]. As with other evidence detailed above, this suggests that salmon can experience negative and positive emotional states.