Memories as Pictures
Kristina Liefke (Ruhr University Bochum)
‘A picture is worth a thousand words’ – ‘Picture this!’ – ‘Picture yourself . . .’ – ‘She sees the big picture’ – ‘He uses artistic license’ – ‘She is painting a black picture (of . . .)’
Many of our familiar idioms about depiction – however trite they may be – capture some salient property of pictures: pictures are informationally rich, can be of particular objects or events, and allow the viewer to take a new perspective or immerse themself in the picture. Pictures can further abstract away from, alter some (or more) details of, or even change the overall valence of the depicted target. In fact, given sufficient creative liberties, pictures need not depict any target whatsoever.
This post investigates the similarities between pictures and the intuitive objects of our episodic memories. It argues that these similarities warrant applying the rich analytical tools from philosophy of depiction to current issues in the philosophy of memory. I contend that such application will help structure and advance discussion in contemporary philosophy of memory.
Pictures and their Properties
Consider the painting below (entitled ‘Drie generaties’ [‘Three generations’]), created 1941–1950 by the Dutch painter Charley Toorop:
Toorop’s painting saliently features a female painter, a man, and a bronze bust. However, its information much exceeds these characters. For one, the painting includes spatial information (e.g., that the characters are inside a windowed room, that the woman is seated in front of the man). For another, it reveals a host of information about the characters and their surroundings (e.g., that the woman is holding a large palette and paint brush, that the bust has oxidized and is larger than life).
Arguably, there is much information about the depicted episode that is not captured in Toorop’s painting. This is due to the painting’s taking a particular perspective on this episode (since the view is from the inside looking out, the picture does not show the back of the painter’s coat), to Toorop’s specific attentional focus (her depiction of the bronze bust is more detailed than her depiction of the coat), and to her idiosyncratic depictive style (note her use of accentuated lines and contrasting colors).
We know from Toorop’s own recounting that ‘Drie generaties’ is a self-portrait, featuring herself, her son Edgar Fernhout, and the bust of her deceased father, the painter Jan Toorop. Since Fernhout did not like to pose for his mother, he would only let himself be portrayed years after Toorop had completed large parts of the painting. As a result, the painting does not depict a single coherent episode.
While ‘Drie generaties’ is reasonably accurate with respect to its protagonists, it deviates from its declared target(s) in some interesting details. Thus, in the painting, Toorop’s face has a surprisingly dark tan – much darker than what one would expect from an inhabitant of the dreary Netherlands (and from her complexion in the photo above). Equally surprisingly, Fernhout – who was likewise a painter, but is not wearing a painter’s coat in the picture – is holding a paint palette.
In a way, Fernhout’s integration into the ‘original’ depicted episode (viz. Toorop painting herself, with her father’s bust in the background) provides an even more striking departure from this episode.
Properties of Memory Objects
The above-discussed properties have all also been observed for the objects of episodic memories. This holds for informational richness (see the literature on mental imagery in memory) as well as for perspectivity and partiality (see the discussion surrounding observer memories), for self-directedness (see the rich literature on autobiographical memory), for aboutness (see Openshaw’s recent SI on reference and remembering), and for the generation of novel content (see phenomena like false recognition and boundary extension).
Just like pictures, the objects of our episodic memories are then representations of a three-dimensional space that can be about a particular object or episode (‘referentiality’), that can correctly represent this object/episode (‘accuracy’), and that can deviate in some details from this episode (‘constructivity’).
Just as memory objects are representations of a past episode from the experiencer’s (original or displaced) perspective, pictures are representations of a situation from the painter’s (actual or imagined) viewpoint.
Recent work on generic memories and on memories from dreams has shown that – like pictures – episodic memories need not be of actual particular objects or events. Rather, they can have non-singular, non-occurrent, and arbitrary referents. These include objects that result from ‘merging’ functionally equivalent objects from different episodes, objects that are imported from other (including counterfactual) episodes, and objects that are supplemented from semantic world knowledge (as evidenced by false recognition cases).
Like pictures, episodic memories have contents that determine their accuracy conditions. Just as “a picture is accurate when the content it expresses fits the target scene it aims at” (where ‘target’ is defined as a perspectival situation; see Greenberg, 2018, p. 865), authenticists about episodic memory assume that a memory is accurate when “the subject’s original experience . . . includes no content that was not included in the subject’s original representation of the event” (Michaelian, 2024, p. 6).
The parallelism between pictures (top) and memory objects (bottom) is graphically captured below:
Since the objects of our episodic memories – unlike pictures – are often not all-visual (field memories typically involve ‘what it felt like’, alongside ‘what it looked like’), the visual nature of pictures seems to point to an inherent limitation of my ‘memories as pictures’-analogy. However, a generalization to other perceptual modalities is also required for the inclusion of multi-sensory and performative art. A related observation holds for the extension from ‘static’ to ‘moving’ pictures, which would be required to maintain the analogy for temporally extended memory objects.
Memories as Pictures
The strong similarities between memories and pictures suggest that analytical tools from one of these areas might be fruitfully applied to the other. This holds especially for the tools from formal philosophy of depiction – see Greenberg’s ‘picture semantics’ – which originate from linguistic semantics, the philosophy of language, logic, and projective geometry. They include truth-conditions, compositionality, model-theoretic interpretation, and geometrical projection.
In particular, picture semantics uses these tools to:
[1] motivate the informational richness of pictures (by showing that its mechanism for representation-to-content conversion yields complex conjunctive propositions),
[2] describe the effect of perspective on pictorial content (by showing how, for the same depicted object or episode, different viewpoints yield different propositions),
[3] explain reference and self-reference in pictures (by providing a careful account of the singular- and centered content of pictures), and
[4] provide a general account of pictorial accuracy that also applies to paintings of counterfactual objects and episodes (by generalizing the depicted target to spatio-temporal parts of possible worlds, i.e. to ‘situations’).
I have recently started to apply these tools to perspective, reference, and accuracy in episodic memory (see this paper, based on my keynote at IPM 4). My initial findings suggest that this application facilitates new answers to several current issues in philosophy of memory. These include:
[1’] a defense of the propositional attitude view of episodic memory (against Sant’Anna, 2018); using the mechanism from (1)),
[2'] a rigorous account of observer memory that distinguishes between field perspective and self-reference (unlike the preceding formal literature; see my 2024, based on (2)),
[3'] an account of non-actual (e.g. dream-) and generic memory objects that gives these objects clear identity criteria and a definite ontological status (as inhabitants of counterfactual worlds respectively as properties; contra Michaelian, 2024), and
[4'] a new argument for pluralism about accuracy concepts and standards (contra Michaelian, 2024).
Along the way, this application supports the compatibility of the propositional attitude view with de se-content, reveals Michaelian’s pisticism (according to which an episodic memory is accurate if it is true of the intentional object of the agent’s experience) as a weak variant of radical authenticism, and explains different intuitions regarding the possibility of observer-perspective memories from dreams.
I am convinced that, at a minimum, the application of picture-semantic tools to the above issues will help (re-)structure the debate in philosophy of memory (e.g., by forcing us to be more specific about established views/positions). The four examples above suggest that this application will even help advance discussion in philosophy of memory, viz. by identifying new or previously neglected positions, and by providing novel arguments for (or against) certain positions.
I expect that picture semantics will have yet many other applications in philosophy of memory, including the continuism/discontinuism debate and issues surrounding reference in misremembering, imagination, and confabulation. I encourage researchers to join in on this exciting project!
