Using statistical tools such as factor analysis and multidimensional scaling, James A. Russell first developed "A Circumplex Model of Affect" in his 1980 paper, wherein he "proposes that all affective states arise from two fundamental neurophysiological systems, one related to valence (a pleasure–displeasure continuum) and the other to arousal, or alertness" (Russell, 1980, as cited in Posner et al., 2005). There have been a number of research papers about the Circumplex Model since then, including "The circumplex model of affect: An integrative approach to affective neuroscience, cognitive development, and psychopathology" by Jonathan Posner, James A. Russell, and Bradley S. Peterson (2005), from which I will draw information for the purposes here. In their paper they say, "Each emotion can be understood as a linear combination of these two dimensions, or as varying degrees of both valence and arousal (see Figure 1)."

"Specific emotions therefore arise out of patterns of activation within these two neurophysiological systems, together with cognitive interpretations and labeling of these core physiological experiences" (Posner et al., 2005). Arousal here is considered to be central nervous system (CNS) arousal, and the valence and arousal systems "presumably are subserved largely by subcortical structures," specifically the mesolimbic and reticular networks (Posner et al., 2005). The signals from these structures are then cognitively interpreted (presumably by the prefrontal cortex) to yield the experience of affective states. 

     There seems to be a problem with the valence dimension in the Circumplex Model. If valence is the degree to which a state feels pleasant or unpleasant, it does not seem that valence can be considered a dimension, since many different states can be equally pleasant or unpleasant under certain circumstances, and these states would be at various angles not aligned with the valence axis. For example, there are situations where excitement can be just as pleasant as happiness, but excitement is not as aligned with the valence dimension as happiness is. If valence is to be a dimension, it should measure something that changes in one direction only. Relative pleasantness-unpleasantness as an axis does not seem to fit the bill, but something like happiness-sadness could. 

    There also seem to be some problems with the arousal dimension. Since pleasantness and unpleasantness can themselves be arousing in their own right, it is unclear how arousal can serve as an axis orthogonal to valence. If arousal is to be a dimension, it should measure something that changes in one direction independently from the perpendicular dimension. It is also troublesome that the arousal dimension is unipolar and does not cross the valence dimension at the origin, meaning the four regions of the affect space are not Cartesian quadrants. An orthogonal bipolar axis in some sense equivalent to arousal could solve these problems. 

     Nevertheless, their statistical methods indicate that these two dimensions (or something like them) produce the circumplex. I believe the dimensions of Pathoscape overcome these problems and significantly improve the model, while remaining consistent with the empirical methods used to develop the Circumplex Model.

REFERENCES

‣ Posner, J., Russell, J. A., & Peterson, B. S. (2005). The circumplex model of affect: an integrative approach to affective neuroscience, cognitive development, and psychopathology. Devolopmental Psychopathology, 17(3):715-734. doi: 10.1017/S0954579405050340. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2367156/

‣ Russell JA. (1980). A circumplex model of affect. Journal of Personality and Social Psychology, 39:1161–1178. https://www.researchgate.net/publication/235361517_A_Circumplex_Model_of_Affect