It's a mistake to place the causal force of pattern recognition in the recognizer. A pattern recognized is a pattern forced. The ingredients at play (components of a recognizer) have no choice but to recognize (as a whole be affected by) the pattern. Fits physicalist idea of math/computation?