The main objective of the TAMED project is to devise new methods and algorithms for realising aspects of general emotional intelligence, one of the core long-term goals of artificial intelligence and artificial psychology. To move towards such an ambitious goal TAMED methods would be required to: a) derive accurate models from small-sized affect data corpora, b) eliminate the subjective biases inherent in affective ground truth, and c) limit overfitting effects of affect models given their context-specific nature. TAMED views general affect modelling from an ordinal perspective and interweaves uniquely novel tensor-based learning models with preference learning approaches. TAMED is highly innovative since it utilises, for the first time, tensor-based and preference (deep) learning models to capture general aspects of affect. Tensor models are characterised by high learning and generalization capacity, and are suitable across different learning paradigms, while preference learning can uniquely eliminate annotation biases and approximate more reliably the underlying ground truth of affect. TAMED methods will be used to investigate the degree to which context-free affect models are possible and general affect patterns can be captured across dissimilar tasks and users. The applicability of the derived models will be tested on the domain of digital games since they offer complex yet well-defined problems for exploring the capacities of general artificial intelligence. The aforementioned innovative aspects make TAMED highly interdisciplinary, with research activities spanning from affective computing and machine learning to digital game design and development. The fellowship will contribute to the researcher’s career development through the acquisition of advanced scientific and technical skills, as well as developing skills within academia and industry. The project will also serve to consolidate and extend the researcher's professional network within Europe and beyond.