Hill climbing is a simple computer science optimization technique. Consider the image below showing a simulation world with a TV (upper left corner), a refrigerator (towards bottom) and some walls.
The blue surface represents entertainment (distance to the TV). The red surface represents the food (distance to the refrigerator). These surfaces were produced with Collaborative Diffusion. A hill climbing agent will look at neighboring values and move toward the one with the largest value. An agent interested in food would climb the red hill, an agent interested in entertainment would climb the blue hill. The hill climbing code to move towards the food could look like this in AgentSheets/AgentCubes: