Even simple neural networks with Hebbian learning can produce surprising emergent behaviors when their inputs and reward systems interact in unexpected ways - I'd be curious to see if the squid develops any quirky preferences or avoidance patterns after extended training.