RF absorption in human body is not linear in terms of both frequency nor location inside body, which is the whole point of making very complex SAR models for specific case studies.
Discussion
You are right. SAR is more complex than simple power density because of tissue resonance, penetration depths, and the non-linear way the body absorbs different frequencies.
My goal with this post wasn't to replace a high-fidelity 3D SAR simulation, but to provide a 'macro' comparison of the primary variables we can control: distance and transmit power.
Even when you account for the complex absorption profiles of brain tissue, the orders of magnitude difference in distance and power between a phone and an AP remains the dominant factor for total exposure.