Following these arguments, Foster & Sjoblom(1996) address quality improvement drivers. They discuss the traditional learning curve of quality improvements in organisations, but broaden the discussion of learning by introducing the distinction
between autonomous and induced learning (Dutton& Thomas, 1984) in production and operations in organisations. They find that up-stream variables such as product design, infrastructure, supplier and customer-related variables are key drivers of quality improvement, and support a much broader perspective than the traditional ‘learning-by-doing’ (the traditional learning curve model), or the autonomous learning, which may be supplemented by induced learning. Ittner et al. (2001) find more support for quality-based learning models that assume learning as a function of both proactive investments in quality improvement and autonomous learning-by-doing,than for models that assume learning as a function of reactive investments in quality improvement alone.They find that benefits from different types of prevention expenditure vary, and that past non-conformance expenditures provide learning opportunities that allow the organisation to cope more efficiently with future failures, thereby reducing subsequent non-conformance costs。