He designed the cams through trial and error to mimic his own gait. Getting it to walk backward was a lot easier, and was the first step, so to speak, since the designer could consciously perceive his own subcomponents of motion while doing a strange act. Walking forward is so far down the neural subsumption stack so as to be difficult to decompose.

This reminds me of Hawkins’ memory-prediction framework for intelligence. Here’s the relevant <a href="http://jurvetson.blogspot.com/2005/01/thanks-for-memory.html">section from my blog</a>:

The 30 billion neurons in the neocortex provide a vast amount of memory that learns a model of the world. These memory-based models continuously make low-level predictions in parallel across all of our senses. We only notice them when a prediction is incorrect. Higher in the hierarchy, we make predictions at higher levels of abstraction (the crux of intelligence, creativity and all that we consider being human), but the structures are fundamentally the same.

More specifically, Hawkins argues that the cortex stores a temporal sequence of patterns in a repeating hierarchy of invariant forms and recalls them auto-associatively. The framework elegantly explains the importance of the broad synaptic connectivity and nested feedback loops seen in the cortex.