In a generic spin-polarized Fermi liquid, the masses of spin-up and spin-down electrons are expected to be different and to depend on the degree of polarization. This expectation is not confirmed by the experiments on two-dimensional heterostructures. We consider a model of an N -fold degenerate electron gas. It is shown that in the large-N limit, the mass is enhanced via a polaronic mechanism of emission/absorption of virtual plasmons. As plasmons are classical collective excitations, the resulting mass does not depend on N , and thus on polarization, to the leading order in 1/N . We evaluate the 1/N corrections and show that they are small even for N = 2.PACS numbers: 71.10.Ay, 71.18.+y, 71.10.Ca The observation of an apparent metal-insulator transition in high-mobility Si metal-oxide-semiconductor-fieldeffect-transistors (MOSFET's) [1] challenged the scaling theory of localization [2], which predicts that a twodimensional (2D) system undergoes only a continuous crossover between weak and strong localization regimes. Although there has been a substantial progress in understanding of transport and thermodynamic properties of MOSFET's and other heterostructures [3,4], the origin of the observed phenomena is still a subject of discussion. Although a conventional (dirty) Fermi-liquid (FL) theory [5,6] can account for many observed effects at least qualitatively and, in some cases, quantitatively, there is also a number of non-FL scenarios for the anomalous metallic state [7,8]. On the experimental side, the main argument for the FL-nature of the metallic state is the observation of quite conventional Shubnikov-de Haas (ShdH) oscillations [3,4], which implies an existence of well-defined quasiparticles albeit with the renormalized effective mass m * and spin susceptibility χ * s . The ShdH and magnetoresistance experiments show that at low densities both m * and χ * s are significantly enhanced compared to their band values [4] and, according to some studies [9,10], even diverge at the resistive transition point.Although none drastically non-FL features of the metallic state have been found in ShdH measurements as of now, there is one very intriguing observation which does seem to present a challenge for the FL theory, at least in its conventional formulation. Namely, in all studies when the spin and orbital degrees of freedom were controlled independently by applying a tilted magnetic field, the effective masses, m * ↑ and m * ↓ , and Dingle temperatures (impurity scattering rates), T D↑ and T D↓ , of spin-up and -down electrons, were found to be almost the same. Moreover, m * in MOSFETs [11,12] was found to be independent of the spin polarization, whereas T D was shown to depend on the polarization only weakly. In n-GaAs, the effective mass was found to depend on the parallel magnetic field [13]; however, this behavior was attributed to the coupling between the in-and out-ofplane degrees of freedom (Stern effect [14]), which is to be expected in systems with wider quantum wells. Given that the Stern effect is...