Anderson proposes a separation between systems and mechanisms. No matter whether the system is constrained to be a dendritic compartment or whether it is extended to encompass all mechanistically relevant parts, there are tools available to describe the respective situation. The mechanistic model does not necessarily consider systems in isolation from the environment or surrounding processes. Even if the system is defined as the dendrite only, factors influencing dendritic processing as well as the embedding of the system in the overall economy, its organization, have to be considered in order to arrive at an understanding of the system’s functioning (Bechtel 2008, pp. 148–150). On the other hand, I would like to argue that we have good reason to extend the boundaries of the system to encompass all the contributing parts. This is a situation in which the original ascription of a function to a system part has to be revised to accommodate new findings. This process is termed reconstituting the phenomena by Bechtel & Richardson (1993). Although direction selectivity was thought to be bound to or even intrinsically generated in SAC dendrites, it turns out that the system can only be understood in combination with other neural elements that vitally contribute to the mechanism in question.
One advantage that Anderson suggests comes with the differentiation of system and mechanism is that mechanistic components can then be set at a different level of organization than the relevant system. The SAC dendrite is at a lower level compared to the input from bipolar cells and the network structure (bipolar cells and neighboring SACs) that enables SAC direction-selectivity. But once the question of how exactly we should carve up the brain into systems and mechanisms has been answered, I don’t think that complex inter-level relationships are much of an issue for mechanistic accounts. They can be easily accommodated within the framework of top-down causation proposed by Craver & Bechtel (2007). They suggest that any reference to inter-level interactions can be analyzed in terms of within-level causal relationships between parts of entities, where parts and entities are related in a constitutive fashion and entities can be located on different levels. Emphasizing the fact that complex inter-level interactions often need to be considered in order to offer adequate explanatory accounts in neuroscience is important, but it is not outside the scope of current models of mechanistic explanation.