This paper presents a novel type of artificial neural network, called neural plasma, which is tailored for classification tasks involving few observations with a large number of variables. Neural plasma learns to adapt its classification confidence by generating artificial training data as a function of its confidence in previous decisions. In contrast to multilayer perceptrons and similar techniques, which are inspired by topological and operational aspects of biological neural networks, neural plasma is motivated by aspects of high-level behavior and reasoning in the presence of uncertainty. The basic principles of the proposed model apply to other supervised learning algorithms that provide explicit classification confidence values. The empirical evaluation of this new technique is based on benchmarking experiments involving data sets from biotechnology that are characterized by the small-n-large-p problem. The presented study exposes a comprehensive methodology and is seen as a first step in exploring different aspects of this methodology.