What does a fish know?

After discussing fish feelings, Jonathan Balcombe moves onto fish cognition: What are fishes’ mental abilities? 

“Here’s an example of fish intelligence, courtesy of the frillfin goby, a small fish of intertidal zones of both eastern and western Atlantic shores. When the tide goes out, frillfins like to stay near shores, nestled in warm, isolated tide pools where they may find lots of tasty tidbits. But tide pools are not always safe havens from danger. Predators such as octopuses or herons may come foraging, and it pays to make a hasty exit. But where is a little fish to go? Frillfin gobies deploy an improbable manoeuver: they leap to a neighbouring pond. 
How do they do it without ending up on the rocks, doomed to die in the sun? 
With prominent eyes, slightly puffy cheeks looking down on a pouting mouth, a rounded tail, and tan-grey-brown blotchy markings along a 3-inch, torpedo-shaped body: the frillfin goby hardly looks like a candidate for the Animal Einstein Olympics. But its brain is an overachiever by any standard. For the little frillfin memorises the topography of the intertidal zone—fixing in its mind the layout of depressions that will form future pools in the rocks at low tide—while swimming over them at high tide!” (Part IV) 

This, Balcombe says, is an example of cognitive mapping. 
Or: 

“… But can fishes spontaneously invent tool use, as we can when unexpected conditions require us to improvise? In May 2014, a study highlighted an example of innovative tool use by Atlantic cods being held in captivity for aquaculture research. Each fish wore a coloured plastic tag affixed to the back near the dorsal fin, which allowed the researchers to identify them individually. The holding tank had a self-feeder activated by a string with a loop at the end, and the fishes soon learned that they could release a morsel of food by swimming up to the loop, grabbing it in their mouth, and pulling on it. 
Apparently by accident, some of the cods discovered that they could activate the feeder by hooking the loop onto their tag and then swimming a short distance away. These clever cods honed their technique through hundreds of ‘tests’—and it became a finely tuned series of goal-directed, coordinated movements. It also demonstrated true refinement, because the innovators were able to grab the pellet a fraction of a second faster than by using their mouth to get the food. That fishes are routinely expected to interact with a foreign device to feed themselves is impressive enough, but that some devised a new way of using their tags shows a fish’s capacity for flexibility and originality.” 

In these chapters, Balcombe talks about lots of studies regarding fishes’ cognition skills. These studies demonstrate that fishes, or at least the fishes used in the experiments, can create mental maps, remember escape routes, have good memory, be trained to perform tricks, learn from experience, innovate to solve a problem, use tools (for example, smash open a clam against a rock), learn from observing other fishes (observational learning skills), and so on. 
They’re smarter than we think. 
He also talks about tigerfishes catching and eating birds, which means that they must know about the distortion of image over water surface and how to calculate to make a leap from the right angle and with the right speed so they can ambush the bird—planning instead of making random leaps and snaps at the air. 
I’m disappointed he doesn’t mention the bird-eating fishes in David Attenborough’s documentary.
At the end of Part IV in What a Fish Knows, Balcombe talks about the plurality and contextuality of intelligence, and our “wobbly criteria for gauging intelligence”. There are, instead, “multiple intelligences”, manifesting in different sets of skills and abilities. If something is critical for a species’ survival, it would be good at it.