Continents slowly drift at the top of the mantle, sometimes colliding, splitting and aggregating. The evolutions of the continent configuration, as well as oceanic plate tectonics, are surface expressions of mantle convection and closely linked to the thermal state of the mantle; however, quantitative studies are so far lacking. In the present study we use 3D spherical numerical simulations with self-consistently generated plates and compositionally and rheologically distinct continents floating at the top of the mantle in order to investigate the feedbacks between continental drift, oceanic plate tectonics and the thermal state of the Earth's mantle, by using different continent configurations ranging from one supercontinent to six small continents. With the presence of a supercontinent we find a strong time-dependence of the oceanic surface heat flow and suboceanic mantle temperature, driven by the generation of new plate boundaries. Very large oceanic plates correlate with periods of hot suboceanic mantle, while the mantle below smaller oceanic plates tends to be colder. Temperature fluctuations of subcontinental mantle are significantly smaller than in oceanic regions and are caused by a time-variable efficiency of thermal insulation of the continental convection cell. With the presence of multiple continents the temperature below individual continents is generally lower than below supercontinent and is more time-dependent, with fluctuations as large as 15% that are caused by continental assembly and dispersal. The periods featuring a hot subcontinental mantle correlate with strong clustering of the continents and periods characterized by cold subcontinental mantle, at which it can even be colder than suboceanic mantle, with a more dispersed continent configuration. Our findings with multiple continents imply that periods of partial melting and strong magmatic activity inside the continents, which may contribute to continental rifting and pronounced growth of continental crust, might be episodic processes related to the supercontinent cycle. Finally, we observe an influence of continents on the wavelength of convection: for a given strength of the lithosphere we observe longer-wavelength flow components, when continents are present. This observation is regardless of the number of continents, but most pronounced for a single supercontinent.