An important distinction to be made in understanding 'pain' is its difference from nociception. In order to experience pain, animals must respond to potentially painful events in a way that shows the animal is not exhibiting a nociceptive reflex (i.e., its changes in behaviour are not a simple reflex response to noxious stimuli, which are physical/chemical/biological stimuli that are harmful or potentially damaging). This "experiencing pain" requires an animal to be able to construct a cognitive/mental representation of the noxious event causing the response. Two criteria for establishing pain perception, which are characteristic of declarative representations, are outlined below:

1) A 'whole-animal' response to the noxious event: behavioural and physiological alterations occur outside of simple reflexes, with long-term responses including avoidance and protective behaviours from the noxious stimulus. These same reactions should also be able to be reduced throug the use of analgesics/painkillers.

2) The pain experience should influence the animal's future behavioural decisions on such an event (e.g., seeking analgesia or paying a cost to reduce its pain, or avoiding the noxious stimulus and learning to avoid future encounters (examples of adaptive responses)).

Behaviours of teleost fish can be adversely affected (as a result of noxious stimuli, such as exposure to venom/acid) to the point where the complex, anomalous responses (such as rubbing of snouts into gravel/tank walls, rocking back and forth, and exhibiting these behaviours for over an hour after exposure) strongly suggest a significant level of discomfort/pain (i.e., more than a reflexive response).

Furthermore, there are examples of teleost fish finding a noxious stimulus (electrical stimulation) so aversive they altered their behaviours to avoid it, but then had reduced avoidance responses to the same stimulus after anaesthetics were administered. Changes in behaviour that occur after the species after exposed potentially painful treatments, which are then reduced by painkillers, strongly suggest that such initial changes in behaviour are a direct result of the 'painful' experience.

There is still the possibility that many of these indicators outlined above (e.g., avoidance behaviours, declarative responses) could be occurring in animals which simply use the same neural pathways that humans use for automatic, unconscious actions programmed to respond adaptively for the simple benefit of the animal's survival.

Many argue instead, however, that consciousness itself is a Darwinian adaptation evolved through natural selection; even the simplest emotional responses to an affective state (e.g., pain or fear) are widespread amongst vertebrates, and this is to be expected considering the great fitness value that comes with such an integrated system (protecting the animal from future harm or death).

Studies have also been done to gain empirical evidence that conscious experiences are present in vertebrates. Distinct neural ensembles in a portion of the brain in rats have been identified to encode the negative affective valence of pain, not simply nociception. It is likely that similar studies could be conducted to demonstrate the same capabilities in other aquatic species, including fish, frogs, and others. 

In the following video Dr Lynne Sneddon presents some of her research into sentience and pain in fishes.

NB: This third-party video is linked for illustration only and copyright and control belong entirely to the originating organisation