The question at the heart of the matter is really instinct versus actual pain. Consider the age old experiment, you accidentally place your hand on a hot stove. Your body will pull that hand back very quickly... then it starts to hurt. The pulling back was a response from information that only went to your spinal cord and back, when the information made the full trip from the brain and back, that is when pain was involved. The pulling back, instinct, only needs a spinal cord, but the actual feeling of pain requires a highly developed brain. Today we know what parts of the brain are the pain receptors, since they light up like a christmas tree on MRIs when slight pain is administered to test subjects.
So then the argument comes down to, physiologically, do fish's brains have the higher development needed to feel actual pain or are they primarily responding to instinct? The question is tossed around the scientific literature, there is a Professor James Rose from Wyoming whose opinion is that fish's brains are not highly developed enough to feel pain. While others, like Professor Moccia from Guelph Canada whose opinion is that the brain is developed enough, argue that newer studies show that the brain is more similar to humans' brains than may have been first thought.
There is evidence to support both sides of the debate, the real problem of course is that the fish cannot tell us, and it would be extremely costly to perform an MRI on a fish's brain while performing these pain experiments. Further clouding the issue is that Prof Rose is a supporter for the fish farming industry, who, of course, would really like to think that fish don't feel pain so that they can continue to cram them into fish farms. I cannot speak for Prof. Moccia, so I will not just guess about his motives. So while there is evidence for both sides, it is very difficult to evaluate fairly with the biases that are evident.
At this point, I think that you have to personally decide on your own, though I also think that eventually a more definite answer will be forthcoming some day when the necessary experiments become far cheaper and easier to perform.