I’m not sure whether structural dissociation is the right model for tulpas; my own model has been that it is more related to the ability to model other people, in the way that if you know a friend very well you can guess roughly what they might answer to things that you would say, up to the point of starting to have conversations with them in your head. Fiction authors who put extensive effort of modeling their characters often develop spontaneous “tulpas” based on their characters, and I haven’t heard of them being any worse off for it. Taylor, Hodges and Kohányi found that while these fiction writers tended to have higher-than-median scores on a test for dissociative experiences, the writers had low scores on the subscales that are particularly diagnostic for dissociative disorders:
The writers also scored higher than general population norms on the Dissociative Experiences Scale. The mean score across all 28 items on the DES in our sample of writers was 18.52 (SD = 16.07), ranging from a minimum of 1.43 to a maximum of 42.14. This mean is significantly higher from the average DES score of 7.8 found in a general population sample of 415 , t(48) = 8.05, p < .001. In fact, the writers’ scores are closer to the average DES score for a sample of 61 schizophrenics (schizophrenic M = 17.7) . Seven of the writers scored at or above 30, a commonly used cutoff for “normal scores” . There was no difference between men’s and women’s overall DES scores in our sample, a finding consistent with results found in other studies of normal populations . With these comparisons, our goal is to highlight the unusually high scores for our writers, not to suggest that they were psychologically unhealthy. Although scores of 30 or above are more common among people with dissociative disorders (such as Dissociative Identity Disorder), scoring in this range does not guarantee that the person has a dissociative disorder, nor does it constitute a diagnosis of a dissociative disorder [27,29]. Looking at the different subscales of the DES, it is clear that our writers deviated from the norm mainly on items related to the absorption and changeability factor of the DES. Average scores on this subscale (M = 26.22, SD = 14.45) were significantly different from scores on the two subscales that are particularly diagnostic for dissociative disorders: derealization and depersonalization subscale (M = 7.84, SD = 7.39) and the amnestic experiences subscale (M = 6.80, SD = 8.30), F(1,48) = 112.49, p < .001. These latter two subscales did not differ from each other, F(1, 48) = .656, p = .42. Seventeen writers scored above 30 on the absorption and changeability scale, whereas only one writer scored above 30 on the derealization and depersonalization scale and only one writer (a different participant) scored above 30 on the amnestic experiences scale.A regression analysis using the IRI subscales (fantasy, empathic concern, perspective taking, and personal distress) and the DES subscales (absorption and changeability, arnnestic experiences, and derealization and depersonalization) to predict overall IIA was run. The overall model was not significant r^2 = .22, F(7, 41) = 1.63, p = .15. However, writers who had higher IIA scores scored higher on the fantasy subscale of IRI, b = .333, t(48) = 2.04, < .05 and marginally lower on the empathic concern subscale, b = -.351, t(48) = −1.82, p < .10 (all betas are standardized). Because not all of the items on the DES are included in one of the three subscales, we also ran a regression model predicting overall IIA from the mean score across DES items. Neither the r^2 nor the standardized beta for total DES scores was significant in this analysis.
That said, I have seen a case where someone made a tulpa with decidedly mixed results, so I agree that it can be risky.
Sorry, didn’t mean to imply that structural dissociation has anything to do with tulpas. I agree that the birthing a tulpa is likely quite different, and I tried to state as much.
Fiction authors who put extensive effort of modeling their characters often develop spontaneous “tulpas” based on their characters
I see the examples in the linked paper, of the characters having independent agency (not sure why the authors call in an illusion), including the characters arguing with the author, even offering opinions outside the fictional framework, like Moriarty in Star Trek TNG, one of the more famous fictional tulpas.
That said, they seem to mix the standard process of writing with the degree of dissociation that results in an independent mind. I dabble in writing, as well, and I can never tell in advance what my characters will do. In a mathematical language, the equations describing the character development are hyperbolic, not elliptic: you can set up an initial value problem, but not a boundary value problem. I don’t think there is much of agency in that, just basic modeling of a character and their world. I know some other writers who write “elliptically,” i.e. they know the rough outline of the story, including the conclusion, and just flesh out the details. I think Eliezer is one of those.
I wonder how often it happens that the character survives past the end of their story and shares the living space in the creator’s mind as an independent entity, like a true tulpa would.