Sich et al. reply:

Regarding the terminology used in our recent publication (Sich, M. et al. Nature Photon. 6, 50–55; 2012), we want to point out that it is known that both the linear and nonlinear parts of the refractive index contribute to soliton formation and are intrinsic optical properties of matter. Understanding the impact of these properties on the propagation of photons, such as dispersion, nonlinear frequency conversion and soliton formation, does not in most cases require a departure from the concept of photons. This is the limit of the 'weak coupling' between light and matter.

The first example of solitons we gave in the introduction of our work is that of solitons in optical fibres, which are pulses of light and are thus readily described in the limit of weak light–matter interaction. We therefore referred to these as 'light-only solitons'. In the case of strong coupling between light and matter, as is realized in the semiconductor microcavity used in our work, we deal with excitation frequencies where the cavity (and therefore light) resonance is absent and the excitonic resonance is also absent, so that neither light nor excitonic (matter) waves exist for the frequencies under consideration. Only half-light, half-matter polaritonic resonance exists at these frequencies, and we therefore refer to these as half-light, half-matter solitons — or 'polariton solitons'.

Investigations of microcavity solitons cited in the correspondence by Barland et al. deal with the case of weak coupling, where the energy levels of photons and material excitations do not hybridize and no polaritonic quasiparticles are formed. As such, in our terminology, this falls into the category of light-only solitons, which, as in optical fibres and other settings, cannot exist without the surrounding matter. In this respect we should also mention that adiabatic or non-adiabatic dynamics of the material excitations is a factor that is secondary to the existence (or non-existence) of polaritonic energy levels and to the choice of operating frequency with respect to these levels. We hope our comments here address the terminology concerns discussed in the correspondence by Barland et al.