When the basic status of a theory is clear, and all that needs to be cleared are details, you can collaborate. But if the main structure of a hypothesis isn't established, and you want to change the paradigm - like it was the case in the 1960s - it's better to work alone.