top of page

Emotions in fish - what is missing for proper welfare legislation?

Updated: Mar 21

The expression of emotions in fish is quantifiable, and they affect the fish’s reality. Emotions make up part of the neurological toolbox, promoting sentience in fish. Advocacy work for fish welfare regulation should be extended significantly to prevent the suffering of billions of fish produced every year.

In humans, welfare is felt consciously and can be verbalized. Dogs can bark, and we can know with some confidence whether they want to go for a walk, beg for food, or feel scared. Fish's social cues are much more subtle, so there is no simple way to discern how fish feel - but we can learn to listen to the ways they make their emotions known.

Evaluating the welfare of fish is a tough task. Fish do not smile, do not cry - they have “poker faces”. Fish live in an environment that is as inhabitable for humans as Mars. We may well be perceived as space invaders by them, as they are somewhat Martians to us. This makes our emotional connection with fish difficult, and it adds to our skepticism about their real capabilities (and of their real suffering).

That is why in fish welfare research, fish were at first considered simplistic, rudimentary, and primitive animals, incapable of any feeling whatsoever. Fish were just a healthy source of protein, synonymous with food. Industry changes were being made to improve fish growth - but they dismissed the conditions in which this growth occurred, including poor welfare!

Fish are not Martians anymore

In the last two decades, in the face of new evidence, scientists recognized that fish can feel both positive and negative experiences and that the balance between the two is important for Darwinian fitness and, therefore, for fish welfare.

The scientists who disagree with a feelings-based approach to fish welfare ascribe a degree of unacceptable anthropomorphism, i.e., the attribution of human traits to fish, to this research. However, the anthropomorphic approach allows for a deep insight into behavioral and physiological emotional reactions and translates them into the human language on behalf of the fish. The fish’s point of view is now encompassed, for the first time, after Charles Darwin’s claim that the differences among species are in degrees and not in kind - including emotions [1].

Given that emotional states are often associated with human psychology, this conclusion may sound surprising at first. Emotions in fish are maybe not as complex and conscious as in humans, but Darwin suggested that the evaluation of animals’ behavior is just the interpretation of their emotional states regarding a specific event or condition, just like for humans.

The study of emotions for Martian-seeming creatures like fish is particularly interesting since they represent a divergent evolutionary branch from tetrapods. But let’s be frank, much more alien are the invertebrates from whom we also ascertain a degree of sentience, like crustaceans and mollusks [2]! From the physiological (and also cognitive/behavioral) perspective, fish are notably similar to other vertebrates. Therefore, the study of emotions in fish provides insight into the evolution of sentience for a “simpler” vertebrate animal that is evolutionarily distant from humans. This is a critical turning point for fish welfare, and might (and should) contribute to the extension of the welfare considerations we currently grant other vertebrates to fish.

100 years late, fish might dream for a seat on the table

Darwin introduced the idea that animals experience emotions analogous to human fear to varying degrees. Every day, research unlocks the mystery regarding the true capabilities of fish and their defiance of aquatic stereotypes. For example, did you know that fish can “sense” through walls? Most cartilaginous fish and over 300 bony fish are equipped with electroreceptors, which enable them to perceive natural electromagnetic fields through vibrations and chemical cues in the water. This allows them to not only detect a wall but also sense what is behind it. Fish have an x-ray ability that humans can only dream of.

The Nobel prize in Economic Sciences winner and philosopher Amartya Sen, in his book The Standard of Living, rhetorically asked: “Why must we reject being vaguely right in favour of being precisely wrong?” Despite the criticism of the anthropomorphic approach, it is seen as a valid starting point. And let’s be fair-minded: the reasonable doubt about the real level of fish sentience is what has led us to discover fish’s mental capabilities and aspire to understand them even more.

Most fish have electroreceptors that allow them to perceive electromagnetic fields. They can sense what is behind a wall in addition to sensing the wall itself.

So to the million-dollar question: Do fish have conscious feelings of which they are aware? Its answer is critical for welfare in a way that would influence our emotional link with them, and advocacy work for fish welfare regulation can be extended significantly through everyday scientific evidence.

Though the breakthroughs in convergent (or divergent) areas of fish welfare like ethology, neurobiology, sociobiology, and ecology have been deciphering what the world looks like to fish, researchers still frequently set aside the issue of whether and in what ways fish might be conscious of their emotions - emotions that even we can only narrowly define.

But let’s see some examples: Many studies have shown that fish have a qualitative experience of the world, can learn and remember, can predict incoming events, have a sense of time, can “draw” mental maps of their environment, and associate time and place [3]. Fish can interact with and recognize their group members, and learn from them [4]. Fish, like primates, innovate, and use tools. These results reveal that fish capacity goes far beyond simple reflexes, instead indicating shifts in mental state. But, can we say that emotions are operating these behaviors?

The view that fish lack conscious feelings is largely based on the fact that fish lack the neural structures that control feelings in mammals. But wouldn't this be similar to concluding that fish cannot breathe because they have no lungs? The fish brain contains the basic structures that lay the groundwork for the more elaborate structures of other vertebrates (such as the hippocampus, amygdala, and cortex areas), and they are also remarkably similar in organization [5].

So again, can we confidently say that emotions are operating these behaviors? Science says yes!

Fish cannot report the subjective experience of feelings, but can rather reflect emotional states.

For many scientists that agree with a feelings-based approach, the mystery is no longer whether fish feel pain, fear, or pessimism, but instead, they investigate the question of which commonalities are shared by fish, including emotions, and how to use this information to improve fish lives. Nevertheless, it is prevalent to confound or overstep fish’s real cognitive capacity (as we do for birds and reptiles). It has been suggested that the emotional “computational” requirements of fish are simpler than what has previously been implied, as it was for birds and reptiles [6]. For our human eyes and our complex mind, this simplicity can deceive our judgment about fish’s real condition, whether physical or psychological.

As humans, we tend to “humanize” everything connected with our own abilities, like emotions - which humans experience as feelings. Fish cannot report the subjective experience of feelings, but can instead reflect emotional states. The assessment of emotional states in fish has to rely on the occurrence of specific behaviors, encoded by their brain states and triggered by the perception of specific stimuli (and its value for the individual at that time).

Fish, like mammals, seek out pleasurable experiences and side-step painful ones.

Therefore, emotional “computational” requirements in fish include the cognitive perception of the stimulus, prompted by the psychological relevance of the stimulus at that point, and then by the assessment of the resources available to cope with it. Decision-making is the last “computational” requirement. It is affected by the individual’s motivational state and sorted into two categories - pleasant or adverse - by psychological evaluation.

As such, it is reasonable to think that, like mammals, fish seek out pleasurable experiences and side-step painful ones as prompted by emotional evaluation.

Fish distinguish between positive and negative stimuli. When given the opportunity, they will avoid painful experiences and seek out pleasurable ones.

Sad fish or happy fish? How can we know?

Humans smile to express joy, nervousness, happiness, or pleasure. Emotions in humans can be felt consciously. They can also be verbalized. In non-human mammals, this is not so straightforward. Many pet owners comprehend their animals’ emotions and how they are expressed by their behaviors, even if the measurement of these emotions is confused or imprecise. But even though owners can only “subjectively assess” their animals' emotional conditions, they act upon the cues given, e.g., by giving them water or taking them to the street.

With no ability to reliably interpret emotions in fish, we have a different way of assessing them: a quantitative assessment. In human research, emotional states, e.g., being relaxed, responsive, content, anxious, or distressed, are the outcome of emotional experiences, underlined by two fundamental dimensions: valence (intrinsic attractiveness or aversiveness of events or experiences) and salience (strength or importance of events or experiences) [7]:

In human research, we use fMRI (functional magnetic resonance imaging) to see which parts of the brain are activated when discriminating between stimuli with different valences and salience. Obviously, we cannot follow this approach with fish. In the study of fish emotions, pain and fear have been the main focus (see here to learn more about this issue). Fish were shown to have and to respect a set of requisites fulfilling the criteria to, as stated by Lynne Sneddon, “characterize pain beyond a reasonable doubt” [8].

In psychological terms, we can clearly see that fish’s behavioral responses to pain are not automatic but rather decided by what it means for them on a case-by-case basis. From these observations, it is reasonable to argue that fish’s pain physiology is similar to that of vertebrates, so why can’t we rely on the vertebrate/human model to evaluate fish?

Fig. 1 Core affect represented in two-dimensional space. Pleasant emotional states are in quadrants Q1, high pleasant intensity, and Q2, low pleasant intensity. Adverse states are in quadrants Q3, low adverse intensity, and Q4, high adverse intensity. Within each quadrant are examples of emotional states.

Alluding to the “Cognitive theory of emotions,” Fig. 1 can help us understand what seems very difficult at the start. This model is used to assess emotional states and moods in humans but has also been applied to “lower-level” animals [9] like bumblebees and fruit flies. Surprisingly, in these animals, findings described emotional states of high intense arousal (Q1 - positive affect, high arousal, and Q4 - negative affect, high arousal) after the individuals’ exposure to an unexpected rewarding event and a repeated adverse stimulus, respectively [10,11]. If this method is used for invertebrates, why not use it to discover the emotional states of fish?

The studies of emotions in fish have been built, however, on behavioral evidence from only one dimension of emotional experience: valence. Studying fish’s perceptions of fear or pain similarly to humans (sensu conscious pain) might give ground for discussion, scientific dissensus, or uncertainty [12]. While the consensus amongst scientists increasingly seems to be that fish can suffer, their preferences, and what exactly makes them suffer is not yet completely understood. This is in part because it is argued that only showing that fish can subjectively feel and discriminate stimuli with different valences and saliences would set forth the path to understanding both fish’s point of view and their emotional amplitude - just like for humans [13].

The ultimate proof of sentience in fish.

Cognitive Biases is the procedure relying on the cognitive theory of emotions, recently used to assess emotional states in fish. It is rooted in the belief that optimistic individuals judge conditions and events they face differently than pessimistic ones.

As we cannot do an fMRI on live fish or ask what they feel, researchers side-stepped the use of imaging technology and designed a way to determine whether fish show specific emotion-like states related to the two core dimensions of emotions. With this, the proof we were looking for to advocate for better welfare regulation for fish is established: Gilthead sea bream (Sparus aurata), a fish species, have been found to exhibit emotions similar to those in humans!

A comprehensive collection of chemical and neural cues of conscious emotion reflected by their inner state was instantaneously decoded by the fish’s behavior, which allowed for envisioning their emotional states [14]. Sea bream were trained in one of four different conditions. They received either a positive event (food delivery) or an adverse event (brief exposure to air) that was either always preceded by a light stimulus (predictable) or occurred at random with respect to a light stimulus (unpredictable). Scientists measured cortisol levels (a physiological stress hormone) and brain activation patterns (neuronal responses) in areas known to be involved in emotional event processing.

Researchers recognized a priori what was at stake and what each condition might mean for themselves if under similar circumstances. The scientists could match the fish’s neurophysiological responses with behavioral patterns that were distinct for each of the four treatments. It was awesome to see that each condition aggregated a set of reactions homologous to humans’ emotional quadrants. This connection allowed scientists to read the emotional state of fish: Pleasant high: excitement/alert - Q1; Pleasant low: calm/relaxation - Q2; Adverse low: depression/sadness - Q3; Adverse high: anxiety/fear - Q4. So, fish prefer predictable and positive events over unpredictable or adverse events, just like us!

A parallel study may put these outcomes in simple words. Female cichlids become sadder and more pessimistic about the world when they lose their chosen mates [15]. And what does this mean? This means that emotional attachment to a partner may not be exclusive to mammals, making fish far more emotionally complex than we give them credit for. If fish can be lovesick, how can we still doubt that they are emotional beings?

These assessments are the most realistic feelings that can be communicated by the fish about the conditions they are being held in, and arguably, this goes beyond any anthropomorphic idealism. The results of these studies do not show equivalence between the fish’s experiences and human emotions, but the emotional states expressed in sea bream are functional, providing the flexibility and speed that ultimately modulates welfare given a specific situation. And do emotional beings not deserve a life without suffering?


Fish are far more emotionally complex than we give them credit for.

We will most likely never understand all there is to know about fish, but research has been filling this knowledge gap. The lack of scientific certainty is not a good reason to postpone measures that prevent real harm to animals, and what we know now is enough to include fish in our moral circle and give them far better treatment than they are currently afforded.


[2] See the review from Broom, D. M. (2007). Cognitive ability and sentience: Which aquatic animals should be protected?, and the report from Advocates for Animals (2005). Cephalopods and decapod crustaceans: their ability to experience pain and suffering, for references.

[3] Bshary, R and Brown, C. (2014). Fish cognition.

[4] Reviewed in Ward, A. J. W. et al. (2020). Social Recognition and Social Attraction in Group-Living Fishes.

[6] Emotional computational requirements are components, such as cognitive, physiological and motivational involved in the formation of emotion. See review in Scherer, K. R. (2009). Emotions are emergent processes: they require a dynamic computational architecture.

[7] Barret, L. F. et al. (2007). The experience of emotion.

[7] Sneddon, L. et al (2014). Defining and assessing animal pain.

[12] See here a recent review outlining the inconsistency and subjectiveness of studies about pain in animals.

[13] Braithwaite, V. and Droege, P. (2016). Why human pain can’t tell us whether fish feel pain.

882 views0 comments


bottom of page