Ibix said:
I think I need to write out the geodesic equations to make sure I followed this.
Well, to be clear, I was talking about the Newtonian limit here. So the metric is time-independent and can be written as diagonal in a suitable coordinate system. Then
[tex]\Gamma^{i}_{0k} = 0[/tex]
(one can always put this term to zero by using a time dependent rotation, as follows from its transformation law), and
[tex]\Gamma^{i}_{jk} = \frac{1}{2}g^{im} [\partial_{j}g_{mk} + \partial_{k}g_{mj} - \partial_{m} g_{jk}][/tex]
We have made a foliation such that we can regard [itex]g_{ij}[/itex] as the metric on spatial hypersurfaces.
This last connection coefficient couples to the spatial velocities. If the spatial curvature perturbations are regarded as "order epsilon" and the spatial velocities also, then this whole term disappears (when you expand around Minkowski; for (a)dS it's a different story of course). Effectively the particle thus only experiences [tex]\Gamma^{i}_{00}[/tex] (it's the only surviving term in the geodesic equation).
This is a subtlety: it doesn't mean that the spatial curvature is zero in the Newtonian limit; it only says that its coupling to the particle's velocity is a higher order epsilon term in your expansion and hence is neglected. Of course, in full fledged Newtonian gravity the spatial curvature
is zero. That corresponds to your metric.