On Perceiving Molecular Time: Computational Chemical Simulations and the Moving Image

Abstract

The perception of time undergoes a radical shift between the human scale and the nanoscale. In an age of rapidly evolving media and scientific technologies, we need to understand how these impact human perception and visual culture. This essay explores computational molecular simulations through the lenses of temporal media theory and moving image practice. Emerging from a creative fellowship with a physical chemistry research group, I focus on two moving image works that depict crystalline structures. One is a nanoscale computational simulation of soot formation and the other is a durational video artwork showing the dissolution of sugar. Computational molecular simulations are shown to produce a feeling of time by smearing an extremely short duration across a longer perceptible duration. This analysis uncovers how the awareness of media as a construct troubles our chronoception (perception of time), while unexpectedly, the screen becomes complicit in scientists’ expert temporal understanding. The videos present vastly different spatial and temporal scales and have different chronoceptive effects: one gives a sense of being within time, the other across time. Ultimately, computational simulations emerge as isomorphic media that have explicit aesthetic properties that connect us to the implicit, abstract energetics of chemical reactivity.

“Time is nature’s way of keeping everything from happening all at once. So the world exists. And the world happens again and again and again. In essence the world is like the different frames of a film reel—a film whose camera view includes the entire universe. There are also, as far as we can tell, an infinite number of frames, infinitesimally separated”.

Andrew Schneider (2015), performance artist

“Nobody, I suppose, could devote many years to the study of chemical kinetics without being deeply conscious of the fascination of time and change: this is something that goes outside science into poetry; but science, subject to the rigid necessity of always seeking closer approximations to the truth itself contains many poetical elements”.

Cyril Hinshelwood (1964), physical chemist

“In the smallest discernible fraction of a second, in the almost instantaneous perception of a sensible quality, there may be trillions of oscillations which repeat themselves. The permanence of a sensible quality consists in this repetition of movements”.

Henri Bergson ([1907] 1975), philosopher

1. Introduction

Because the human mind evolved to perceive time based on the comparatively slow movement of human-scale objects, the devastatingly fast movements of molecular phenomena are difficult to temporally perceive and comprehend. The representation of phenomena, such as electrons in orbit around an atom’s nucleus, historically relied upon static probability diagrams, or illustrations. These depict either impossibly solid objects, areas of possibility, or graphical forms that abstractly imply motion. My first visual memory of a carbon molecule was graphene—a sheet of carbon atoms covalently bound together in a single lattice of perfect hexagons. It was shown to us first as diagrams (see Figure 1). Later, we learned about graphite, the multilayered variant of graphene, using three-dimensional (3D) analog models (see Figure 2).

As relatively stable materials, graphene and graphite do well enough with static representation; however, in order to understand dynamic processes, practical experimentation is paired with computational simulations. These simulations can be thought of as incredibly precise and tunable animations that generatively emerge from physical–chemical systems defined by computational scientists. As a filmmaker and moving image artist working with the Curtin Carbon Group, I have been researching complex representations of the dynamic movement of molecules. I am exploring both scientific and media tools and the properties of their temporal representation. I previously calculated the astounding discrepancy between the capability of the human visual system and the frenetic pace of an oscillating water molecule. If the molecule “vibrates 102 trillion times in one second, and if it were played back at a perceivable limit of ten vibrations per second, it would take 347,856 years to watch one second worth of movement” (Rassell 2019, p. 210). I saw the simulated animations made by the Carbon Group and experienced their movement as extremely rapid, while understanding that these were in fact ultra-slow motion. Over time, this led to the following question: How do molecular simulations mediate our chronoception (perception of time)?

This essay unpacks the relationship between the operative, perceptual, and temporal essences of molecular simulations. Media archeologist Wolfgang Ernst states that “Every technology that succeeds in reshaping the present with its own temporality is a chronotechnology” (Ernst 2016, p. 123). I will explore the ways molecular simulations perform this reshaping, using an analysis of time-critical properties within molecular simulations, the moving image, and media art. The analysis will suggest how media temporalities intervene in the chronoception of dynamic phenomena. Molecular simulations contribute to a transformation of durational perception through radical discretization, to borrow Ernst’s term, as they are arguably some of the most temporally amplified media available1. My research seeks to extract time-theoretical inspirations from the high-tech relations between computational molecular simulations within a media epistemological context. Software studies and electronic engineering are in-depth with regard to the mathematical and technical components of these systems; however, this article goes further to get closer to the temporal essence produced by scientific media. While the experiences and concepts discussed here emerge from practical experimentation with the simulations in an artistic context (i.e., in the production of a moving image installation), this article concerns the foundational characterization of computational simulations in relation to contemporary creative moving image works.

2. Philosophies of Time

Contemporary media philosophy provides sophisticated analyses of the conditions within which modern technical, analytical, and digital media operate and impact upon cultural experiences and understandings of time. The benefit of media philosophy is the way it builds upon the issues dealt with by earlier philosophers of time to analyze emerging media, its materiality, and technical operativity through a cultural lens. The philosophy of time has been entwined with scientific thinking, especially developments in physics, since the time of Parmenides, Heraclitus, and Zeno. Zeno’s paradox of the flying arrow, in which an arrow appears to be standing still at each moment of precise observation, suggested motion was impossible. Where the pre-Socratic philosopher Heraclitus, and following him Henri Bergson, looked to conscious experience to support claims that everything is in flux (becoming), Parmenides and later Immanuel Kant believed in the concepts of timelessness and the continuum (being). Parmenides and Kant wrote about the subjectivity of time; their work claimed that happenings in time are illusory and subjective. Further, the founding scientific doctrine of causal determinism supported the holistic idea of everything existing at once and the arrow of time being an illusory construct of the human mind. In contrast, the theory of infinity supported the ideas of integers and discrete timepoints (Reichenbach [1956] 1999), which favored the concept of an incremental creeping forward in the direction of the arrow of time. In the digital age, the discretization of events or moments of time into frames and then pixels, produced or read linearly, has been segmenting movement into smaller and smaller units of time through the use of technology such as high-speed videography (currently around one million fps) or scientific visualization strategies. Some highly constrained scientific applications, such as laser interferometry, can produce videos of trillions of frames per second; however, it is important to note that this is not a video of the same photon of light but a reconstruction of frames captured of different photons of light (Velten et al. 2013).

Many modern scientific techniques rely on temporally engineered simulations at precise timepoints with exquisitely high temporal resolution. The discretization of time is key to chronoception. A rapidly increasing temporal resolution and evermore complex forms of algorithmically generated movement pose a challenge to our media literacy and intuitive understanding of dynamic phenomena and how we should interpret them or understand their cultural implications (a single frame of a computational simulation is a complex black box of its own). Temporal media theory has described and analyzed how media have been discretized and the ensuing perceptual ramifications of the shifts from film to video to digital moving images. Cinema 1: The Movement-Image (Deleuze [1986] 2013) and Cinema 2: The Time-Image (Deleuze [1989] 1997) saw Gilles Deleuze’s continuation of the aforementioned philosophers’ considerations of being, becoming, holism, and segmentation. In these texts, he reconceptualized Bergson to explore the interplay between discrete and continuous temporality in cinema. This investigation allowed us to reimagine time differently and our understanding of cinema’s temporal richness. The scientific revolution, for Deleuze, comprised the relating of movement not to “privileged instants” (a single, important image), but to “any-instant-whatever” (a regular, equidistant series of images, some important, others not so important) (Deleuze [1986] 2013). Practically, this concept has been pushed to its limits via the technical discretization of events, e.g., via high-speed photography and video and the digital segmentation of individual frames into pixels.

Media archeologists use three distinct categories that are helpful in considering sense perception and our feeling of time: First, technical media are a direct recording of physical senses, tapes, and film photography (Barker 2018), and with these we undergo what postphenomenology refers to as “analogue mediation” (Ihde 1979). In this case, our sense experience is directly indexical via technology to the stimuli, e.g., when light transmission is visual throughout photographic transmission. Second, analytic media breaks a perceived whole into parts, e.g., the chronophotography developed by Étienne-Jules Marey beginning in the 1880s (Rassell 2022). In terms of human chronoception, analytical media seem to privilege the perspective of the instrument. Third, digital media rely upon discrete signal processing that allows for “translation mediation”, i.e., they translate one sensory modality into another and thus collapse all senses into one media (Kittler 2013). Ironically, this collapse is exactly what allows for the reengineering of information into a multisensory experience and allows the return to a dynamic and embodied connection with the world.

3. Simulating the Formation of Graphenic Materials

Graphite has many industrial applications, e.g., in lithium ion batteries, and so its optimal formation as a pure and conductive material is a subject of interest among materials scientists. In Tami Spector’s exploration of temporal chemical aesthetics, she discusses the historical neglect of transitional states between two stable molecular structures in a chemical reaction in favor of a focus on structure (2017). The transitional or intermediary structures are so ephemeral, having a lifetime on the femtosecond scale, that they cannot easily be experimentally isolated (Spector 2017, p. 5). The Curtin Carbon Group has been researching classes of complex defects that occur during the formation of graphite. Although it is common to study the graphitization process experimentally, the Carbon Group is known for their advanced atomistic simulations that depict graphitization as a dynamic molecular process. The resultant 3D models can be viewed statically, by excising one timestep (a single frame)2 of interest, or they can be viewed together as a video to see a structure forming across time. These simulations enable scientists to study atomistic transformations that occur on the femtosecond timescale, significantly smaller than the timescale of physical experimentation, which for graphite is instrumentally limited to seconds with current furnace technology (Jason Fogg, personal communication, 9 February 2024)3.

Computational simulations of graphitization rely upon the variation in multiple parameters, i.e., pressure, temperature, duration of the experimental time, number of atoms, volume, and initial geometry. Computationally, this process is similar to stop-motion animation in that one timestep is calculated before moving onto the next “frame”. The aim is to capture an improbable “event”, e.g., the seed of a deformation in the structure that grows over time and leads to a defective material. Crucially, when examining the chronoception of molecular simulation, instead of being indexically and temporally linked to an experimental process of a real event, there is no “real time” because the event is virtual and its speed is controlled by a numeric code that stands for temperature. The higher the temperature defined in the simulation parameters, the more quickly the reaction will proceed, and the more that happens between timesteps. This allows a shorter experimental time to capture structures of interest, but also risks the event moving too rapidly to be correctly described.

Temperature and time are thus intimately interlinked. The systems set up for these simulations become more ordered as heat is removed. In physics, the directionality of the arrow of time (past to present to future) has been defined by the dissipation of energy. Entropy increases a forward movement in time; the cooling cup of coffee, an entropic process, cannot be reversed, and the arrow of time proceeds into the future. Structure then emerges from the system as energy is dissipated. Sometimes, in the formation of a molecular structure, there are no discernible changes for a time, followed by a frenetic series of changes. At a constant timestep rate, some early computational simulations of carbon structures took a year to calculate but only captured meaningful events over a comparatively short duration. Nigel Marks developed an algorithm that allows for automated scaling of the rate of timesteps (Marks and Robinson 2015). Using predictions from increasing energetics in the model, this strategy involves a concertina effect on the temporal resolution. A timestep of 0.2 fs as the energetics in the structure increase, implying an impending rapid cascade of change in the molecular structure, is followed by an increase in the timestep as the energetics slow and the changes in the structure decrease4. In essence, to capture an event, the simulation must have a balance between the duration of the experiment and the timestep (the temporal resolution).

A second approach to time-ramping simulations is a hybrid simulation method. Martin et al.’s video (Martin et al. 2021; see Figure 3) shows two graphene molecules interacting within a flame over 41.6 picoseconds. The production of soot is an example of graphitization. Martin et al. created a hybrid simulation by merging two computational modeling approaches—Classic Molecular Dynamics (MD) and Quantum Mechanics/Molecular Mechanics (QMMM)—in order to increase the temporal and spatial detail around a specific event, i.e., the formation of the first bond between two layers of graphene. Typically, MD is a method for analyzing the physical movements of molecules and atoms, while QMMM allows the modeling of the electronic arrangements involved in creating chemical bonds. The motion of electrons is much faster (10⁻¹⁸ s or 1 attosecond) than the movement of atomic nuclei, and this comparison means scientists can algorithmically decouple the motion of the electrons from the motion of the atomic nuclei. To re-couple and synchronize MD and QMMM, you need a 0.2 fs timestep, and the simulation for this complex hybrid model was developed by Laura Pascazio (Martin et al. 2021).

The video contains one discernible shift in playback speed as it changes from MD to QMMM. The duration of video produced via QMMM captures one thousand times (three orders of magnitude) more temporal resolution than MD when femtoseconds per second of video are compared. Our perceptual experience of the simulation is such that we move from perceiving a “rapid” speed to a speed that seems slow enough that we can perceive the details of motion that are occurring. That is, until the new bond forms extremely quickly (see the red bond in Figure 3). This is the temporal nature of the formation of a chemical bond that creates the need for strategies such as variable timesteps and also impacts and is impacted by issues of aesthetic and temporal representation. In the soot simulation, the single change in speed (due to the timestep rather than the playback framerate) provides a perceivable shift of the phenomena, but also of the mathematical technique, each with their own temporo-perceptual complexities. The choice to include a timecode, something that has not yet been standardized in scientific visualizations or simulations in the same way that scale bars have been in microscopic images, gives the viewer a discursive framework of the timescales at play in this simulation.

The energetics in a molecular simulation rely upon the electrons and atomic nuclei in a structure. Hermann Weyl related the energetics of electrons to their temporal mode: “All the physical characteristics of the ultimate elementary particles of matter, particularly of electrons, can be read off from the neighboring field. … Then there is no point like Now and also no exact earlier and later. … The immediate present is not entirely abrupt; there is a small halo, quickly fading toward the past and toward the future, along with the self-shining light of immediacy” (quoted in Ernst (2016), p. 129). This speed of “fading toward the past” in femtosecond precision computational simulations expands atomic time into a dynamic representation in the present. With analytic still images, like Marey’s chronophotographs, an event is “sucked” into them, and the image is “thick with time” (Barker 2018, p. 93). In contrast, the molecular simulation produces a new event that smears across duration such that we may observe it, thus becoming a processual representation. Heraclitus formulated a concept of existence as absolute movement, meaning the world never flows through the same spatiotemporal situation twice. This idea endured in the process philosophy of Alfred North Whitehead. Whitehead said that there is always something happening, before there is something (Whitehead et al. 1979), and for him, the vocabulary of science and philosophy fell short in describing the world of process. Neither are our media, as Barker pointed out, excelling at dealing with such a processual, dynamic world (Barker 2018). Nowhere is this truer than in the media’s attempts to represent the ephemeral, skittering, buzzing, oscillating, processes of the nanoscale. Simulation media, like experimental apparatuses, must arrest a phenomena, store and sort it, and then analyze it (Barker 2018). In this very arrest, the discretization of media, and by extension of time, is complicit in the conceptual reframing of the process as a phenomenon that can be captured.

However, the temporal mode of dynamic computational simulations differ from the analytic image, which is self-referential as a conflation of discrete time-points. For Barker, Marey’s analytic chronophotographs depict a “cultural technique of archiving usually fleeting moments and transducing them into points” (Barker 2018, pp. 84–85). In contrast, the smearing of an exquisitely short duration across a longer duration that occurs in the computational simulations might be thought of as a cultural technique of temporally amplifying the ephemeral event and transducing the invisible energetics of the system into frames and points within those frames. A computational simulation ultimately creates a series of sampled instants, documenting a simulated event that can then be both viewed and analyzed. It creates what might be termed a computationally generated nano-temporality.

In Creative Evolution, (Bergson [1907] 1975) identified the discrepancy between dynamic motion and sensory perception: “In the smallest discernible fraction of a second, in the almost instantaneous perception of a sensible quality, there may be trillions of oscillations which repeat themselves. The permanence of a sensible quality consists in this repetition of movements”. Computational simulations extract, expand, and perform these movements to make timescales sensible that otherwise are not. The temporal “magnification” impacts upon the phenomenological experience of the video. In the case of electronic media (television, for example), Ernst argues that it “circumvents the cultural competence to be able to distinguish unambiguously between historical time and the present”. (Ernst 2016, p. 151). Electronic media have the ability to represent the past as present, as in, here in time. The viewer may know this is illusory but can suspend their disbelief in order to feel co-present with the content unfolding on screen before them. In contrast, a viewer of a molecular simulation will likely be aware that they are observing an artificial construct. The aesthetics are akin to an animation, and most people would expect spatial and temporal limitations to perceiving molecular dynamism. Having a lack of a direct sensorial comparison, i.e., not having seen molecules moving in the world around us with an unaugmented sensory system, makes us aware of the extent of construction required to present a video such as the soot simulation. Although a viewer’s understanding of how much the speed of movement is accurate—relative to real time—is based upon their level of intellectual, or discursive, knowledge.

Carbon scientist Jacob Martin has described that he developed an expert chronoperception of whether the calculations in a simulation are correct based on his visual experience of molecular motion. The first aspect of this is the visual vibrational frequency of a molecular structure’s chemical bonds. The atoms vibrate around each other based on simple harmonic motions on the order of 10¹⁵ hertz, and Martin has a sense of what this speed looks like. Sudden jumps, or the “flying ice cube effect” (where all the points begin to drift in one direction), indicate an issue with the underlying physics—virtual heat can build up due to mathematical errors that cause these types of visual inconsistencies (personal communication, 1 February 2024). The second aspect is the speed of movement of an atom or molecule across the screen. Scientists know approximately how fast phenomena move at the nanoscale, so when something “whips” across the screen, Martin knows that it has traveled a certain distance. Ernst notes that while the screen is a perceptible frame of spatial limitation, the temporal properties are imperceptible (Ernst 2016). However, in this instance, Martin is describing a screen that is complicit as a relative spatiotemporal measurement device. These two properties of chronoceptual experience are the ways a scientist can look at a simulation and intuitively understand the precision of their calculations. This understanding is based upon the representation of mathematics and physics that drive the simulation of molecular structures.

This expertise is developed even in the case of the soot video, which is a digital simulation without an original. The molecular simulation has no analog—no indexical nor formal comparisons can be made. However, the molecular simulations are an isomorphic representation of the complex mathematics and physics at play in the generating algorithms. The ball and stick forms that are chosen to represent change, movement, and dynamics are explicit forms necessary to perceive the implicit mathematics of dynamics, almost a poetic gauze through which we might glimpse the abstract nature of the energetics of the system. The moving image, meanwhile, has proffered experimental and experiential explorations into chronoception.

4. Chronoception in the Moving Image

Since its inception, the moving image has held time as a central concern. The exploration of time-lapse to compress time is exemplified in the work of moving image artists Chris Welsby and Emily Richardson, while others such as Peter Gidal and Tacita Dean illustrate the expressive use of deceleration techniques such as slow-motion and long duration. In their films, David Hall and Michael Snow saw time as a sculptural material that had its own specific weight (Elwes 2015, p. 118). Later works created a real-time mapping of the moving image installation time to the viewer’s time—Douglas Gordon’s 24 Hour Psycho (Gordon 1995) was a 24 hour slow-motion screening of Alfred Hitchcock’s Psycho (1960), and Christian Marclay’s The Clock (Marclay 2010) charted every minute of the day in real-time through the editing together of cinematic depictions of clocks. Kevin Jerome Everson’s Park Lanes (Everson 2015), an unedited eight hour observation of a factory workday, and Andy Warhol and John Palmer’s Empire (Warhol and Palmer 1965), a 485 minute slow-motion film of the Empire State building as day turns into night, are additional examples of the durational style, i.e., works that focus on the passage of a long period of time. Moving image theory, for example, Catherine Elwes’ Installation and the Moving Image (2015) and Deleuze’s aforementioned Cinema 1 ([1986] 2013) and Cinema 2 ([1989] 1997), have scrutinized the moving image and cinema and their temporal turns.

Our understanding of the human visual system, which encompasses the eye, retina, and visual cortex, has contributed to the engineering of time with media technologies being developed in relation to physiological operations and limitations. A well-known example is the standardization of film at 24 frames per second (fps) due to being the lowest rate that consistently gives the illusion of movement to a human viewer. With dynamic phenomena such as reactive chemical structures, a high framerate is necessary so the media can then be slowed to perceive detailed movement, rather than just a blur. This framerate is determined by the relationship between the speed of movement and the frequency limitation of the human visual system. A second example of a technological design that relied upon creating an illusion for the human visual apparatus is the Nipkow disk. A human eye retains information for 100 ms, and this duration informed the engineering of a rotating analog image scanning device that was developed for mechanical televisions. The Nipkow disk had holes at specific intervals and could complete a full frame of an image in under 100 ms. This design avoided any glitches that would otherwise occur through the perception of sections of different frames at the same time. These two examples illustrate physiological properties that provide constant, somewhat objective effects on the human visual system while other aspects of chronoception have been shown to be subjective. The fields of psychology and neuroscience have found chronoception to be flexible—time interacts with our bodies, our memories and the world around us (Black 2018). Stress, memory, repeated behaviors, expertise, and enjoyment have all been shown to impact our chronoception.

Chronoception is accompanied by a raft of media concepts that impact the operative representation of dynamic phenomena. The associated concepts of presence, real-time, latency, delay, memory, immediacy, and instantaneity have been studied in depth, particularly within the field of media archeology (see, for example, Wolfgang Ernst, Friedrich Kittler, Timothy Barker, and Sean Cubitt). Multiple technological developments have shifted cultural notions of past, present, and future. The ability to record an event via photography and the temporal engineering of analog and later digital media have repeatedly upended the notion of past, present, and future. Paul Virilio saw this triad as being “surreptitiously replaced by two tenses, real time and delayed time, this relative difference between them reconstitutes a new real generation” (Virilio 1994, pp. 53–54). Ernst, in his treatise Chronopoetics: The Temporal Being and Operativity of Technical Media (Ernst 2016), would later label this shift a new “Zeitreal” (temporal reality).

Artists working with molecular and quantum phenomena have created audiovisual works that play on the quirks of the human mind to represent complex temporal phenomena in experimental ways, thus contributing to the zeitreal. Media artist Ryoji Ikeda’s immersive installations often push the viewer’s perception to the limit. An extension of the performance Superposition (Ikeda 2012), the installation Supersymmetry (Ikeda 2014) arose from Ikeda’s residency at the European Organization for Nuclear Research (CERN) in Switzerland. Immersed in a vast dark gallery, the viewer wanders between lightboxes blinking beneath swarms of particles that accumulate and disperse, then through into a gauntlet of parallel screens that materialize and vanish. Data are rendered as particles and then assigned numbers and text for what seems like mere milliseconds, only to be obliterated, leaving the illusory afterimage of trails of light. The installation makes use of rhythmic clicks and synthetic beeps that skitter across the eardrum in an ephemeral and tactile manner. These brief moments of sound are barely distinguishable from one another, what Curtis Roads (2001) would classify as micro sounds—those that are delimited by the threshold of human perception (2004). Ikeda’s treatment of quantum phenomena saturates the sensory system with its multiple screens, flickering frames, and sonic frequencies that, at times, approach sensory overload, only to dissipate into moments of absence. Temporally, Ikeda seems less interested in providing a slowed-down version to allow the viewer to wholly grasp any form of phenomena; rather, he overwhelms the viewer with pulsating, there-one-moment-gone-the-next, media.

Paul Thomas’ Nanoessence (Thomas 2010) is an interactive projection of nanoscientific graphs that were layered in the Unity real-time game development platform. Unity was used to create a fly-through that the viewer controls through a customized breath interface. Nanoessence requires the viewer to breathe in order to navigate through the layered nanoscape of skin cells. In Thomas’ work, there is a limiting of temporal experience, that is, a range of speeds within which the interface functions. However, the control of temporal interaction with the nanophenomena is handed to the viewer via the inbuilt stop–start mechanism of the breath interface. Black (2018) suggests that interactivity “represents an attempt to introduce a human temporal scale into the time of the machine, rather than a mechanic temporal scale into the time of the human, producing not an economy of speed but an economy of waiting” (Black 2018, p. 147). The interactivity in Nanoessence gives the viewer the opportunity to increase or decrease the amount of stimuli, perhaps for them to catch up with or come to terms with the strange phenomena they are encountering. I suggest this type of interactive interface, which gives the viewer the agency to control the amount of information being perceived across time, is an emerging stage in the evolution of media that present dynamic molecular structures.

5. Sweet Apparition: Video Art as Durational Observation

Richard Frater’s Sweet Apparition (Frater 2008) is a 19 min silent, durational video that shows a structure built from white sugar cubes slowly dissolving inside a fish tank. Karen Donnellan’s Arc (Wax and Wane) (Donnellan 2016–2017) depicts glass melting in real time; notably, this is one of few other examples of creative video that depict crystalline structures in a process of reformation—in this case from an ordered crystalline structure towards a disordered amorphic form. In Frater’s work, water is poured into the tank, and over time, the middle portions of each cube are the first sections to dissolve, leaving an intricate lace-like lattice—reminiscent of the diagrammatic crystalline structures in the previous scientific visualizations—that slowly folds to the ground as the structure dissolves. Although the dissolution begins quickly, and we see crystals of sugar falling rapidly from the structure (see Figure 4), and the structures have largely collapsed by the two-minute mark (see Figure 5), the dissolution of the skeleton remains mesmerizing, eventually floating to the top of the water bath (see Figure 6) in a kind of reverse collapse. The denouement of the work is not the final detachment of the sugar structure from the bottom of the tank but when movement stops occurring.

Like any durational work, Sweet Apparition embodies two modes of duration: the mode of the phenomena being observed and the mode of the viewer’s perception. The latter is what Hamish Win calls the viewer’s “tenure of endurance” (Win 2020), and this is linked to a subjective chronoception. What this kind of work achieves, and how it goes beyond scientific visualizations in the representation of physical and chemical interactions, is that it connects the viewer to a perception of duration. This sense of duration is missing in the static representations in Figure 1 and Figure 2 of graphene and graphite, respectively, although it may exist alongside a wooly understanding of molecular interactivity as something devastatingly fast and incomprehensible to the human visual system. The durational experience, banal and evocative of stasis, even in its dynamic videographic medium as it may be, directs our attention to the activity of the sugar molecules, making us aware of the movement and temporality at the molecular scale that we cannot otherwise visually perceive and instilling an awareness in the static nature of what we have been presented with in other locales (i.e., science communication and scientific diagrams).

Sweet Apparition is observational in that it relies upon a fixed camera position. There are no rapid cuts or swift camera moves—cinematic techniques that help us to imbue a subject with dynamism and speed. For Deleuze, “natural [visual] perception introduces halts, moorings, fixed points or separated points of view, moving bodies or even distinct vehicles,” i.e., it is subjective, whilst cinematic perception, that of the camera, “works continuously, in a single movement whose very halts are an integral part of it and are only a vibration on to itself”. (Deleuze [1986] 2013, p. 26). As with slow cinema, the languid shot encourages the viewer to create their own subjective gaze path across the image rather than conforming by returning to the center of the screen with each successive cut (Coutrot et al. 2012). As far as viewer perception is concerned, the engineering of time is at its most minimal, thus allowing the viewer’s subjective chronoception to be activated. In comparison to the soot simulation, with its duration of 00:37 s, Sweet Apparition draws the viewer’s attention to the act of observing the human-scale structural outcome of a chemical process over time.

If the soot simulation intricately and imperceptibly segments time in order to extend it for human perception, then Sweet Apparition privileges the perception of duration over an engineering of time that blinkers us to the “privileged instant”. Black asked, “why do different kinds of [media] objects seem capable of producing different kinds of time?” (Black 2018, p. 167), what we might chronoceptively articulate as feelings of time. Because simulated structures are computationally constructed, they have no reference to real-time phenomena and are therefore untethered in time—their temporal properties are constructions, which are virtual. The soot simulation is a discretized video that obscures the perception of real-time duration, as (Bergson [1907] 1975) claimed of the act of segmenting movement. However, the flow of data in this simulation-as-digital-video is not just a flow that points towards the future (due to the predictive qualities of the simulation’s generating algorithm) but is also a flow of information that is designed to extend the event. This extension occurs via an extreme form of computational slow-motion that has the effect of unfolding the past into a present reality. Conceiving the simulation in this way is to erode its discrete nature, to shift it into the realm of the whole, and this in turn impacts the chronoception of the viewer. The holistic real time is not in contrast to, but complementary to, the discretized, smeared perception of time. Watching Sweet Apparition, we have the sense of being across time; with the soot simulation, we are in time.

6. Conclusions

There is a chronoceptual disjunct between the viewer’s experience of highly dynamic motion and the “fact” that they are observing a billionth of a second over several seconds. Discursive, numerical text-based timecodes, however, are perceptually unhelpful, and the techniques of media arts and the moving image, such as embodied light and sound (Ikeda), may be helpful in phenomenologically initiating people into a femtosecond timescale of molecular computational simulations. Other techniques used by moving image practitioners, from durational video (Frater), to sophisticated creative coding (Ikeda) and interactive interfaces (Thomas), can offer additional strategies for working with computational simulations and 3D models.

This characterization of science and moving image media and chronoception has drawn upon practical creative experimentation with producing molecular simulations, conversations with scientists, and a temporal media analysis of moving images from physical chemistry and creative media. This investigation has elucidated the following. Molecular simulations smear an event of infinitesimally short duration across a temporally perceptible duration. This smearing, or temporal amplification, of a short duration allows the viewer to feel co-present in time with the motion occurring on screen. The chronoception of a viewer will be influenced by memory, expertise, and discursive knowledge of the phenomena being depicted, although a reasonable viewer would be expected to understand that this is not a real-time or live action video, in part due to the animated aesthetics. Molecular simulations are reflexively aware of their own status as temporally generated media, i.e., as simulations. The dimensions of the screen unexpectedly become a spatiotemporal measurement tool in the viewing of simulations by expert scientists, providing a spatial reference for the structure’s movement. This supports the suggestion that chronoception is heavily reliant on memory and prior experience. I found the chronoception of computational molecular simulations to be isomorphic; that is, the visual and temporal aesthetics are explicit forms of the implicit physics and energetics that are used by the simulation algorithms in order to generate dynamic media. Analyzing the soot simulation alongside Sweet Apparition, two videos with vastly different timescales, allows us to compare the feeling of being within the time of nanoscale phenomena in the case of the soot simulation and across time in the instance of the durational Sweet Apparition.

Advancing computational simulation strategies such as those described here have dynamic temporal compression, both technically and chronoceptually—i.e., the rate of movement ramps up and down. Variable time-ramping and hybrid approaches such as the QMMM/MD discussed in the soot simulation analysis are evidence that computational simulations are beginning to share temporal engineering strategies that are extensively researched in the moving image studies and cinema studies. Temporal compression is often perceptually precise in narrative film; that is, a viewer has a real-time reference to compare the media to. In contrast, with computational simulations of molecular phenomena in motion, the viewer has no real-time reference. Where an expert has a highly developed conceptual frame of reference for perceiving the dynamics of computational simulation videos, i.e., the vibration of a bond in picoseconds, a visitor to an art gallery may have an entirely different temporal frame of reference. In this gallery context, appealing to the viewer’s cinematic literacy and techniques of media arts may contribute to a more intuitive sense or embodied perception of the physics at play within chemical interactivity.

A better understanding of the temporal attributes and impacts of media and scientific techniques leads to dialogue and pragmatic flow between two aligned practices. The project that is emerging from this characterization of carbon simulations is the development of custom-built software that combines, plays, and time-ramps stereoscopic, human-scale, live-action videos with 3D computational simulations. The resulting software, and the experimental documentary that is being produced using this system, will provide opportunities for testing temporal media philosophy and theories of chronoception. Immersive 3D experiences add the options of giving the viewer temporal control and 360-degree sound. Additionally, emerging media tools such as the artificial intelligence (AI) tools AlphaFold and Sora will impact our cultural conception and perception of temporality. AlphaFold is an AI tool for the prediction of molecular structures that draws upon protein databases constructed over decades. Sora is a text-to-video generator based on a large learning model that can produce up to one minute of video. The implication of these types of AI tools and their time-remixing abilities on molecular dynamics have yet to be studied in depth. Machine learning is already used to more accurately describe interactions between atoms, but it has not yet been used to reproduce molecular dynamics. An exploration of this area would further expand the interdisciplinary study of the technical, cultural, and perceptual implications of emerging scientific media.