Peatlands cover 3% of the world's land area and store approximately 30% of the world's soil carbon, thus playing an important role in regional and global carbon cycling. Northern peatlands are particularly vulnerable to climate change, which has motivated research on the response of peatlands to allogenic forcing. However, less attention has been given to the self-regulating potential of peatlands. In this thesis, I describe peatlands under the conceptual framework of complex adaptive systems, wherein cross-scale feedbacks can potentially dampen or accelerate peatland response to disturbances. I hypothesized that the strength of feedbacks, and therefore the self-regulating potential of peatlands, decreases with increasing allogenic forcing. To test this hypothesis, I studied the feedbacks between ecosystem structure and function at different thaw stages in the Stordalen peatland in northern Sweden; a peatland in the discontinuous permafrost zone, where active thaw is occurring. I also collected data at the more stable and autogenic Mer Bleue peatland in eastern Canada. My results support that there are strong feedback mechanisms between structural variables (vegetation, water table and microtopography) at Mer Bleue (Chapter 3). I investigated carbon function (litter decomposition and CO2 and CH4 flux) and its link to structure at Stordalen (Chapters 4 & 5) and found that litter decomposition has weak feedbacks with structural variables along the permafrost thaw gradient (Chapter 4). To further disentangle the structure-function feedbacks, I studied 10 thaw stages and found that the dominant structural controls on CO2 and CH4 fluxes vary with progressing thaw (Chapter 5). A comparison of structural feedbacks at Mer Bleue with those at Stordalen confirmed that feedbacks are weaker in the latter more allogenically controlled site (Chapter 6).This research contributes to the bodies of literature on ecosystem self-regulation and structure-function variability in the discontinuous permafrost zone. The key contributions are as follows: 1) Deductive tests of self-regulation in peatlands are rare and my results support the hypothesis that ecosystem feedbacks are stronger in a stable peatland than in a peatland in transition. 2) My findings highlight the role of transitional stages of thaw on the carbon function of permafrost peatlands. Both contributions emphasize the importance of including self-regulating feedbacks and transitional thaw dynamics to better constrain carbon loss from thawing permafrost peatlands in a rapidly changing global carbon cycle.