Von Helmholtz holds that every perceptual experience consists in the above- mentioned components, but the proportion in which these components occur may vary. For example, in a brightly sunlit room perception is characterized by vivid visual sensations, hence it has no element whatsoever that does not derive from an actual direct sensation. In the evening, as the light decreases to twilight, visual sensations are limited to the brighter objects, so the dimly visible objects are seen with the increased contribution of the representations stored in the memory of previous acquaintance with the furniture. In complete darkness, the perception enables one to find one’s way about in the room without hitting objects solely by means of representations. In everyday experience it is possible to have a smooth transition between the sensory and the representational contribution to perception, but the inferential process must always be constant. Even in a brightly lit room, perception cannot take place without it. For instance, the spatial properties of a table are perceived if a representation is formed from several views of the table from various sides and distances as a subject moves, which can be integrated with the sensations obtained by touching its surfaces. Besides, the perception of the table is accurate only if it enables the subject to deduce correctly the expected sensations were her eyes and hands brought in a determinate relation with the table.
The question of space exemplifies how the analysis of particular problems of research depends on the theoretical decomposition of perception. It is indeed consistent with von Helmholtz’s epistemology that space perception does not represent existing relations among physical objects. The perceptual spatial properties are symbolic shorthand for the interaction between the sense organs and the physical world. They cannot provide certainty about the existence of spatial relations and the spatial order of things, or even about the space itself. The space experienced in perception is only the form or order that has proven adequate to represent relations among subjects and objects accurately enough for practical reasons, that is, it is the arrangement of external causes that is inductively reckoned by subjects to best fit their practical commerce with the environment (1878,  1925, ill: 206-207; Hatfield, 1990; Turner, 1993).
The aggregates of sensations of spatial perception are the data tied together as long as the coordinated movements are carried out. Von Helmholtz (1878) calls “presentabilia” the collection of the possible sensations that give rise to aggregates if subjects execute particular voluntary movements in any given time. The knowledge base is the association between the motor impulses and the voluntary movements that is learned by taking the changes induced on the coordinated aggregates of sensations into account. For each voluntary movement subjects feel the nervous impulse imparted on the muscles and observe the correlated change of sensations. By predicting which sensations will occur for given impulses and inferring which changes new impulses may yield, the subjects learn by trial and error the association between impulses and movements.
The spatial value of sensations depends on a “non-sensory” quality that von Helmholtz calls a “local sign,” by which the eye position is coupled with a visual direction, that is, the imaginary line of the foveal sight, requested by the intention to fixate a point in the visual field. Once the visual direction for the fovea is mapped, the other retinal points are localized according to their mapping on the visual field in relation to the fovea. This means that for each position of the stationary eye, a map of voluntary motor impulses onto visual directions allows one to associate any retinal point with the movement required to let the eye fixate the field point that stimulates it by reflecting light. The local signs endow the retina with continuous positional values and the mapping constitutes the reference system for eye movements such as abduction, elevation or rotation (von Helmholtz,  1925, iii; 1878; Lenoir, 1993; Hatfield, 1990). The local signs act as shorthand for the motor impulses required to make visual field points pass from the retinal periphery to its center, to make sensations appear or vanish when movements are performed or reversed, to localize sensations in the field. Then the perceptual space is the projection of this mapping by subjects who represent this learned coupling to themselves as the outward distribution of visual or tactual sensations. Thus the subjects can learn that if some movements repeatedly make an aggregate of sensations occur, while the reverse movements make it disappear, the elements of the aggregate are stationary as long as movements are carried out. The aggregates among presentabilia that occur in different times even for the same voluntary movements are identified with what turns out to be the things of the outer world. If, furthermore, in the memory representations, the knowledge is stored that this aggregate occurs at will any time a sequence of movement is performed, then subjects may infer that this aggregate corresponds to the object that is the plausible independent cause of sensations.
Even the spatial ordering of sensations is constructed by means of a projection that depends also on the nature of the sense organs. For instance, if fingers are moved over a surface, the same sequence of tactile sensations arises regardless of the finger that one may use. To obtain the same sensations at will, there is no need to repeat the same forward or backward sequence of movements. The sensations do not follow one another in a fixed sequence and the motor impulses are different from those required to move fingers over points in a row. For tactile aggregates do not follow a linear order, subjects infer a higher-order layout so that presentabilia are arranged adjacently to one another. Furthermore, since it is possible to reach each point from any other through different classes of motor impulses, giving rise nevertheless to the same aggregate, the adjacent order takes the form of a surface. If, then, distinct surfaces require different classes of equivalent motor impulses, subjects project sensations into a three-dimensional layout because this is the layout that contains the tactile aggregates presented for the required movements.
Source: Calì Carmelo (2017), Phenomenology of Perception: Theories and Experimental Evidence, Brill.