Recent empirical studies have reported evidence that many aquatic species, including fish, cephalopods and crustaceans, have the capacity for nociception and pain, and that their welfare should be taken into consideration. Some sceptics, rejecting the precautionary principle, have denied that any study demonstrates pain or other aspects of sentience in fish. This target article discusses some of the scientific shortcomings of these critiques through a detailed analysis of a study exploring nociception and analgesia in larval zebrafish.

• MGS score significantly increases following vasectomy in CBA mice. • Buprenorphine (0.05 mg/kg) does not prevent pain associated changes in behaviour. • Baseline MGS scores in male CBA mice are not zero.

Key (2016) affirms that we do not know how the fish brain processes pain but denies — because fish lack a human-like cortex — that fish can feel pain. He affirms that birds, like fish, have a singly-laminated cortex and that the structure of the bird brain is quite different from that of the human brain, yet he does not deny that birds can feel pain. In this commentary we describe how Key cites studies that substantiate mammalian pain but discounts the same kind of data as evidence of fish pain. We suggest that Key's interpretations are illogical, do not reflect the published empirical evidence for pain in fish, and are out of touch with current thinking on brain evolution. We agree that more scientific data are needed to understand how the brains of fish — and other organisms — function. But the incompleteness of current knowledge certainly does not constitute evidence for inferring that fish in particular do not feel pain.

The majority of commentaries are supportive of our position on the scepticism that muddies the waters surrounding fish pain and sentience. There is substantial empirical evidence for pain in fish. Animals’ experience of pain cannot be compared to artificial intelligence (AI) because AI can only mimic responses to nociceptive input on the basis of human observations and programming. Accepting that fish are sentient would not be detrimental to the industries reliant on fish. A more proactive discussion between scientists and stakeholders is needed to improve fish welfare for the benefit of all.