When pattern recognition is used without adequate knowledge and reasoning (e.g. incomplete patterns without recognition of common variations or subcategories) the risk of error is high. Clinical patterns should be learned but not relied on. They should not be rigid, rather they should be continually tested and revised. Having fixed patterns can lead to errors and hinder learning variations or sub-categories. That is, if fixed patterns are all you look for, you miss variations and don’t learn new patterns. Poor use of pattern recognition can lead to “centring”, where pattern identification is biased by a key feature and competing patterns (i.e. differential diagnosis) are not tested for.