"Humans are agents capable of helping others, learning new behaviours and forgetting old ones. The evolutionary approach to archaeological systems has therefore been hampered by the 'modern synthesis' - a gene-centred model of evolution as a process that eliminates those that cannot handle stress. The result has been a form of environmental determinism that explains human evolution in terms of heroic struggles and selective winnowing. Biologists committed to the modern synthesis have either dismissed agency as a delusion wrought in our bodies by natural selection, or imposed a sharp, Cartesian split between 'natural' and 'artificial' ecologies. We revisit the seminal literature of evolutionary biology and show that the paradigmatic fault lines of 21st century anthropology can be traced back to the 19th century and beyond. Lamarck had developed a two-factor evolutionary theory - one factor an endogenous tendency to become more advanced and complex, the other an exogenous constraint that drove organisms into conformity with environment. Darwin tried to eliminate the progressive tendency and imposed linearity constraints on evolution that Thomas Henry Huxley rejected. When experimental evidence falsified Darwin's linear hypothesis, the race began to develop a new, gene-centred model of evolution. This became the modern synthesis. The modern synthesis is now under pressure from the evidence of anthropology, sociology, palaeontology, ecology and genetics. An 'extended synthesis' is emerging. If evolution is adequately summarised by the aphorism survival of the fittest, then 'fitness' cannot always be defined in the heroic sense of 'better able to compete and reproduce'. The fittest organisms are often those that evade selective winnowing, even when their ability to compete and reproduce has been compromised by their genes. Characteristically human traits like language, abstraction, compassion and altruism may have arisen as coping strategies that allowed genetically vulnerable populations to negotiate new ways of being fit. The extended synthesis allows for the possibility that great apes were agents long before they were human and that this agency enabled them to fit their environments to their own needs. This article summarises features of the extended synthesis that seem most relevant to archaeology. Some of the topics it discusses may seem abstruse and perhaps unnecessary because they amount to an acknowledgement of socio-natural complexities archaeologists have understood for decades. However, they are extremely significant in study-domains where biology and archaeology intersect. Archaeologists can no longer uncritically accept the conclusions drawn by molecular geneticists because the theoretical framework of evolutionary biology is under reconstruction."