THE PATTERN
Figures that
will not hold still.
What keeps returning across the paintings, counted rather than asserted. Method, numbers and the places the numbers fail are all stated.
A single painting can be read any way at all. A body of work cannot be read that freely, because a pattern either shows up across it or it does not. So the archive was put through a description pass, work by work, against a fixed list of markers, and the results were counted.
This page reports what that count produced. It also reports the part where the method proved unreliable, because a page that hides that is worth less than one that does not.
What was done
A language model was shown each painting image and asked to assign a phenomenon type, an intensity score from zero to ten, and to describe the image using the marker vocabulary. Nothing was scored from memory or from a title. The description was of the image itself.
The largest run covered 873 entries. Seven were file errors, where the image could not be found or opened. The remaining 866 paintings each returned a usable description, and those 866 are the basis of everything below.
The types
Each painting was given one type, and the type names come from the vocabulary of sleep. The largest group, a little over a quarter of the works, were described as hypnagogic: images arriving at the threshold of sleep. Close behind them, around a quarter, were described as hallucinatory. Around one in six were described as dissociative, where the sense of the self or the scene coming apart is the subject. A handful were described as hypnopompic, the waking edge.
And roughly one painting in three was judged to carry no such phenomenon at all. That figure belongs in the result as much as the others do. It is the reason this can be called a pattern rather than a selection, because a third of the archive was left out of it.
The markers
Separately from the type, each image description was searched for marker families, and a painting counts once for a family however often the family appears.
Dissolution, of edge, ground or form, is the most common thing in the whole archive: it appears in more than half the works. A boundary treated as a problem, rather than simply drawn, appears in close to half. Figure against ground, in around two works in five. Named hypnagogic imagery, and named hallucinatory imagery, each in about a third. Luminous or glowing passages, in close to a third. Collapse of form or identity, in more than a quarter. Phantom or half-realised figures, in close to a quarter. Chimeric or fused forms, in about one in seven. A figure read into a surface, the kind of seeing that turns a wall into a face, in about one in eight.
Those families are not scattered evenly across the work. Around six paintings in ten carry at least one of them. Around three in four carry two or more at once, and in the same painting.
That last figure is the one worth sitting with. It says the concerns are not occasional here. They overlap, and they overlap in the same works, which is what a fixation looks like rather than a mood.
Where the method fails
An earlier pass scored about three hundred paintings against the same taxonomy, under a different file-naming convention. Twenty titles turned up in both runs. Those twenty are the only place where the same painting was scored twice, and they show what the method is worth.
Of the twenty, ten were given a lower score the second time and four a higher one. Six kept the same score but twelve changed type label. The movement was not random. Every painting that was called hallucinatory in the first run and changed in the second changed to dissociative, never the other way. That is a sign of a definition that shifted between passes, not of a second look at the art.
The conclusion is narrow and it is meant to be. The type labels and the marker counts are usable as description. The intensity score is not reported here as a finding at all, because on the repeat run it moved on most of the paintings it touched. It is left out rather than averaged, and the exact tables behind this page are held on the record with the sources.
What the count does not say
It does not say the paintings are dreams. A description pass over a painting, made by a model, is a reading of a painting. It is not the artist's testimony and it is not a clinical observation.
It does not say the imagery is rare. Painting is full of dissolving ground and figures that will not settle. What the numbers establish is narrower and more useful: that these concerns are not occasional in this body of work. They are the constant.
The one part of the chain that is genuinely unusual is not on this page. It is the clinical record, which is set out on Waking and dreaming.
Where to go next
Waking and dreaming, the clinical record →