Design Challenges and Opportunities of Fossil Preparation Tools and Methods

Fossil preparation is the activity of processing paleontological specimens for research and exhibition. Alongside traditional mechanical extraction methods, fossil preparation presently comprises non-destructive digital techniques within the emerging field of virtual paleontology. Despite significant technological advances, traditional and digital preparation remain cumbersome and time-consuming. However, this field has received scarce attention from a human-computer interaction standpoint. This study aims to present the current state of paleontological fossil preparation, highlighting its key challenges and initiating a dialogue about innovative designs to address current issues. Our research comprises a qualitative study involving technical preparators and virtual paleontologists. The study consists of two main parts: Firstly, a focus group session brought together preparators and researchers to discuss their workflows, obtain a preliminary understanding of their issues, and ideate solutions based on their counterparts’ experiences. Subsequently, a series of contextual inquiries involved direct observations and semi-structured in-depth interviews. By transcribing and analyzing the data through theoretical and inductive thematic analysis, we clustered emerging themes and applied concepts from human-computer interaction and related fields. Our findings report on challenges traditional and digital fossil preparators encounter, shedding light on potential opportunities to enhance their tools and workflows. This work contributes a novel analysis of fossil preparation from an HCI perspective.


INTRODUCTION
Paleontology can be defined as "the study of the history of life on Earth as based on fossils" [44].This scientific field sheds light on evolutionary processes, morphological changes, and humanity's place among other forms of life on Earth [12].Fossil preparation, one of paleontology's primary endeavors, is described as the activity of "preparing fossil specimens for use in paleontological research and exhibition" [53].One of its main activities is the meticulous removal of the sediment matrix surrounding fossil specimens, the organisms, fragments, and traces of organisms preserved in strata [7].Technical preparators are the professionals responsible for the fossil preparation process [6].Whybrow et al. (1985) describe preparators as specialists who possess a thorough knowledge of anatomy, considerable manual dexterity, an innovative mindset, and mechanical or chemical skills [51].
Technological advances enabled fossil preparation to evolve through the novel methods of Virtual Paleontology, which can be defined as "the study of fossils through three-dimensional digital visualizations" [48].This field is becoming increasingly relevant due to the new possibilities offered by the non-destructive exploration of digital fossils, which is particularly important for studying unique and fragile fossils [8].Instead of mechanically removing fossils from surrounding matrixes, modern-day paleontology researchers perform a process named Digital Preparation, in which they segment fossil data into distinct regions of interest [1].Digital methods enable scientists to analyze the internal structures of tiny fossils, an investigation that would require damaging processing under traditional methods.Besides internal morphology, Virtual Paleontology enables researchers to process fossil data to create 3D models that can be employed in functional analyses [35].Such models allow researchers to simulate how extinct species interacted with their environment, and they can also be used for scientific communication and educational purposes [27].
Despite the importance of analog and digital fossil preparation, there have been limited studies concerning the optimal design of tools that take into account the specific challenges of this discipline.
In traditional fossil preparation, some of the tools that are commonly used in mechanical extraction were adapted from other fields and have failed to address the specific needs of technical preparators [28].Virtual paleontology incurs similar issues as this new discipline has emerged from the adaptation of medical imaging methods that do not fully accommodate the requirements of this use case [9].For example, automated segmentation algorithms employed in medicine are seldom applicable to paleontological datasets, and the manual exploration of fossils through current interfaces has proven limitations [8] [40].In fact, digital preparation tasks such as semi-automated image segmentation are acknowledged as a tedious, time-consuming, and error-prone process [55].
Motivated by the aforementioned issues, we sought to broaden our understanding of paleontology's state-of-the-art practices and artifacts in order to identify opportunities to enhance existing tools and to inform the design of future technologies better suited to this field's specific challenges.Indeed, we conducted focus groups and contextual inquiries with both traditional fossil preparators and virtual paleontology researchers [52].Five experts participated in our focus group, followed by contextual inquiries involving seven paleontologists.We visited paleontological labs across Germany to conduct direct observations of different defossilization procedures and conducted in-depth interviews with experts.Our observations were recorded with photos and videos demonstrating their mechanical extraction techniques.Moreover, we documented the workflows of researchers employing virtual paleontology methods to identify issues through observational and think-aloud methods [45].We analyzed our transcriptions and images through theoretical and inductive thematic analyses [3].Our classification organized the data according to emerging themes, existing task taxonomy, and theories in human-computer interaction and related fields.The analysis enabled us to identify the limitations of current paleontological tools and evidenced opportunities for the design of new tools that would be more suitable to the needs of fossil preparators.We incorporate insights from various cognitive, perception, and human-computer interaction theories in order to formulate hypotheses regarding the potential benefits of novel interaction techniques in fossil preparation.In summary, this paper discusses results from qualitative studies with paleontology experts and proposes design implications of future technologies for effectively processing paleontological data and specimens.

BACKGROUND
A large body of research investigates fossil preparation techniques and artifacts.Literature on this subject dates from Fritz-Gaertner's The Preparation of Rocks and Fossils for Microscopical Examination in 1878 [13].Other notable early publications include Bather's Preparation and Preservation of Fossils in 1908 and Hermann's Modern Laboratory Methods in Vertebrate Palaeontology from 1909 [2] [15].In a contemporary survey, Whybrow et al. present fossil collection and preparation techniques from a historical perspective [51].
Regarding Virtual Paleontology, Cunningham et al. provide an overview of this field from a historical and natural sciences perspective.[8].Racicot et al. focus on fossil scanning and its applications in paleontology research [37].Pandolfi et al. provide an updated overview of recent developments in Virtual and Computational Paleontology [34].Abel et al. provide palaeobiologists with a comprehensive guide to digital fossil preparation techniques [1].
Concerning the taxonomy of fossil preparation tasks, Leiggi et al. offer a comprehensive guide of tools and methods involved in "specimen collection, preparation, conservation, and reproduction" [28].Karim et al. focus on a description of digitization workflows for paleontology collections [19].Buccella et al. created a taxonomy of paleontological activities from a Geographic Information Systems (GIS) perspective, focusing on software engineering and Software Product Line (SPL) development [5].
Preparators have also been the subject of qualitative studies reporting on their tools and methods.Wylie et al. conducted ethnographic analyses with several preparators to produce an essay and a book that illustrate "the elements of technique, science, and art involved in the multifaceted work of a preparator" [53] [54].Shever et al. report on anthropological fieldwork investigating the complex and unpredictable process of "discovering and reconstructing prehistoric nature" [42].
Although paleontology and fossil preparation tools and methods have been abundantly documented in the scientific literature, there is still a significant gap in investigating this field through the humancomputer interaction lens.Such an investigation is important for the design of novel tools that take into consideration the specific challenges of fossil preparation.

METHODOLOGY
We conducted a qualitative study leveraging ethnographic methods divided into two phases.In the first phase, we organized a focus group with five participants stemming from paleontology research and technical preparation [22].Although an in-person meeting would have been a preferred data collection modality for this event, we faced limitations due to pandemic restrictions, which prompted us to hold this event online.Our focus group was intended to create a discussion between paleontology researchers and technical preparators around their respective fossil processing workflows.Participants described their current tools and methods and discussed current challenges as well as opportunities for improvement.We merged these two distinct groups of professionals so that they could learn from their counterparts' workflows and potentially ideate novel solutions inspired by such knowledge.Since both paleontology research and technical preparation involve several processes, we focused on one aspect where these two groups overlap -the fossil preparation process, specifically concerning the separation of objects of interest from the surrounding matter.In this one aspect of their endeavors, both professional groups have the mission of optimally extracting fossils, whether the specimens are physical or digital.
In the second phase of our study, we conducted contextual inquiries involving direct observation and semi-structured in-depth interviews [52].These inquiries aimed to expand the knowledge acquired in our focus group, gaining further insights on preparation tools and workflows and obtaining feedback on the concepts generated in the study's previous phase.Our contextual inquiries involved on-site visits, whenever possible, and online meetings when this modality sufficed to study digital workflows.In this phase, we interviewed and observed seven participants.We recorded our focus group session in an audiovisual format.Contextual inquiries were recorded in audio, and certain passages of our direct observation were filmed and photographed.We transcribed our audio and video materials and selected significant excerpts for coding.
The data analysis procedure followed the approach described by Braun and Clarke [3].This process was performed by two researchers: Author 1, who is a human-computer interaction doctoral student, and Author 2, who is a professor in the same field.Author 1 performed a preliminary analysis of the transcripts to familiarize himself with the data contents.Next, this researcher created an initial list of potential codes using inductive coding as a data-driven approach for recurrent subjects, as well as deductive coding for concepts derived from human-computer interaction, psychology, neuroscience, and paleontology literature.Deductive coding comprised both theoretical frameworks and existing task taxonomy.Next, Author 1 grouped codes into potential themes.Then, Author 1 presented a list of codes and themes to obtain feedback from Author 2, who has more significant experience in qualitative methods.The two researchers discussed themes and codes and ideated further theories and concepts that could potentially capture themes in the data as codes.These discussions allowed the authors to develop a mutually-understandable list of codes iteratively.Once the codebook was completed, Author 1 extracted relevant passages of the transcript as individual quotes and analyzed them using the aforementioned codes.Author 2 analyzed a sample of these quotes to ensure inter-coder reliability.Next, the authors selected a number of themes that they considered relevant to inform HCI practitioners about addressable challenges and opportunities of fossil preparators.Finally, the authors reported their findings as a thematic analysis that contextualizes analyzed data with relevant literature.Findings have been organized into subsections that the authors estimate to flow logically from an HCI perspective.
Our contribution does not aim to offer a comprehensive account of all aspects of paleontology, as this wide field spreads across numerous ramifications.For example, our scope does not include paleontology activities such as fossil conservation, taxonomic classification, or collection registration [31][30] [20].Instead, we focus on analog and digital fossil preparation, especially on their respective tasks related to sedimental matrix removal and image segmentation.

FINDINGS
Our findings are organized by the relevant themes identified in our data.Subsections reflect the combination of theoretical and inductive approaches in our thematic analysis.In units where we applied a theoretical thematic analysis, we present constructs that could be discerned from participant responses, although often not directly communicated.Whenever appropriate, we discuss potential opportunities for human-computer interaction interventions that could tackle the challenges presented in our results.Some of the concepts presented in this section arose from discussions in our focus group session and responses in our semi-structured interviews.Other propositions were inspired by existing literature artifacts that originally targeted fields whose tasks share some similarity with our use case.One important rationale for presenting challenges and opportunities in the same subsection space is to make it easier for readers to connect these pieces of information.
Two focus group participants accepted our invitation for contextual inquiries.For consistency, results refer to them under the same participant number across both events.Thus, although five paleontologists participated in our focus group, and seven professionals took part in our contextual inquiries, we refer to their statements using participant codes P1 to P10.

Task Analysis
In our contextual inquiries, we observed participant workflows and listened to their descriptions of tools and tasks involved in their trades.We conducted a task analysis to promote a better understanding of their tools and methods, which is essential to inform the design of new artifacts [39].This analysis enables a better understanding of the upcoming subsections, which make reference to a number of tasks.This analysis is not supposed to be comprehensive, as it is limited by the sample size of paleontology professionals and the types of fossils they process.
Technical preparators can engage in a variety of tasks, including fossil extraction, preparation, conservation, cataloging, and presentation.In this analysis, we focus on fossil preparation as the common link with the work of virtual paleontology researchers.Our inquiries indicate that preparation involves the tasks of Matrix Removal, Stabilization, Restoration, and Finishing.Matrix Removal is the process of removing the sediment matrix in which fossils are embedded [4].We observed participants P1 and P8 cleaning fossils with manual tools such as needles, picks, and brushes, as well as with pneumatic tools such as air scribes and abrasives (Figure 1).P1, P4, and P8 also described the use of chemical acids for matrix dissolving as an alternative that was not used in their respective labs.Stabilization can be done prior to, during, and after the matrix removal process to prevent damage to fossils preserved in substrates that deteriorate under certain conditions [53].P1 and P8 also mentioned Restoration tasks of repairing and reconstructing fossils using supporting materials.P1 and P5 detailed Finishing tasks involving careful retouching for the aesthetic and anatomically correct presentation of fossils for research and exhibition purposes.
For data obtained from Virtual Paleontologists, we applied and extended the taxonomy created by Laha et al. for tasks related to visual analysis of volume data [25].Digital fossil preparation's main endeavor is Image Segmentation, "the process of identifying and delineating objects in images" [49].In order to segment image datasets, P2, P6, P7, and P10 engage in tasks such as Search, Pattern Recognition, and Selection.These participants often manually select regions of interest (ROI) by performing tasks such as Path Following and Filtering on two-dimensional slices.P2 and P10 described Reorientation of image datasets as an important prerequisite to segmentation.Participants also mentioned Navigation, Spatial Understanding, and View Manipulation as supporting tasks for segmentation.After a fossil dataset is segmented, researchers perform a number of tasks depending on their research goals.Quantitative Estimation of scalar data properties is performed by P10 for "evaluating the buoyancy capabilities of specimens, " whereas P2 employs it to obtain "information on [bone] biomechanics" for "functional morphology analysis, " and P6 emulates fossil structures "to perform buoyancy calculations".Another approach to analyzing fossil preparation tasks is through the lens of Activity theory, a descriptive framework that describes activities as purposeful, transforming, and developing interactions between subjects and objects [29].In its application as an HCI framework, Nardi states that activities consist of several actions or chains of actions, which in turn consist of operations [32].Fossil preparation can be considered as an action within paleontology.Each of its operations follows a set of conditions that relate to immediate goals and their respective motives.For example, P1 describes the challenging condition of extracting fossils without creating any scratches on their surfaces.As one of the many goals of paleontology is to establish the causa mortis of specimens for a number of research motives, the preparation action must have fossil integrity as a goal.

Fossil Tomographic Data
Both technical preparators and virtual paleontologists frequently mentioned tomographic data as assisting artifacts and working substrates, respectively.3D characterization techniques such as computed tomography and magnetic resonance enable the nondestructive exploration of paleontological specimens [8].P5 stated that "computed tomography is very important to show the small details of fossils, " while P6 highlighted its need for locating bones and "developing the strategy for technical preparation," and P8 stated that such information allows preparators to decide whether a finding's contents are actually worth their processing time.
Although tomographic imaging is undeniably useful for fossil preparation, several participants reported challenges related to the low attenuation contrast of findings composed of homogeneous materials.P1 reported the limited utility of low-contrast tomographic imaging, such as ammonite scans, which often cannot reveal the specimen's shape and location like its vertebrate counterparts.In this example, a specimen might comprise calcite material and be preserved in calcareous sediment, which would cause its scanning output to yield insignificant contrast [8].P2 stated that such a scenario precludes researchers from utilizing filtering techniques such as thresholding to separate materials, which forces researchers to execute a significant amount of manual work, as exemplified in Figure 2. P3 added that even manual processing might be unfeasible in certain cases where a researcher "cannot see quite clearly where the bone ends".P7 revealed that paleontology researchers working with ammonites often favor working with rare hollow specimens due to the low contrast of more common ammonites filled with sediment of similar radiodensity.P6 stated that "hollow fossils are quite rare," demonstrating the impact of low-contrast datasets on the digital preparation of most scanned specimens.Participants also reported issues caused by the absence of tomographic imaging of the fossil to be processed.P1, P8, and P10 stated that the majority of technical preparators do not have access to the tomographic images of the very piece they are processing.P1 added that such resources are reserved for findings that are potentially important or rare, and quite often, she resorts to online searches to envision their next steps in preparing a specimen.The absence of tomographic data is impactful to technical preparators as such information is vital to prevent unnecessary work on findings that might ultimately be fruitless.For example, P8 reported discovering that a fossil was "destroyed in the middle" and beyond repair only after spending a significant amount of time working on the specimen.The participant expressed his wish for such data as a means to assess whether certain fossils warrant the lengthy preparation work.
Although HCI cannot influence the material properties of fossils or the low contrast of their scanned output, there are opportunities for improvement for the interactive systems used in digital fossil preparation.As attenuation contrast might be below users' discrimination threshold, designers could potentially present alternatives for data rendering [33].For example, leveraging multisensory integration might improve the perception of contrast and understanding of complex structures, particularly in low-contrast data [46].Incorporating haptic rendering into image visualization would provide users with tactile feedback on features within fossil volumes.Such additional information could enhance visual data rendering and improve the detection of subtle changes in data properties that may be visually indistinguishable [14].Haptic feedback could be especially helpful in communicating boundaries between different material structures.Auditory feedback may also be leveraged to provide paleontology researchers with more information about a low-contrast scan.For instance, different frequencies could be associated with density levels or other data characteristics in synchronization with the cursor or segmentation tool as it moves over different fossil structures, making it easier to detect subtle density changes.

Mental Imagery
One salient theme that emerged from our inquiries is the reliance that both fossil preparators and paleontology researchers have on mental imagery.Mental imagery is a type of subjective experience that resembles the perception of objects, scenes, or events in the absence of direct external stimuli [21].Imagery allows humans to activate and manipulate internal mental representations to complete tasks that require knowledge regarding absent input [38].Both physical and digital fossil preparation workflows leverage mental imagery.P1 compares fossil preparation to completing a puzzleuncovered parts provide cues she mentally completes based on existing knowledge, leveraging imagery to guide her cutting process.P1 and P4 reported the strategy of searching for uncovered fossil corners as cues for recognizing embedded fossils, i.e., matching the stimulus to a mental representation of previously experienced fossils.P8 highlighted the importance of experience, which, over time, allows preparators to recognize fossils by observing minimal parts of them.P3 reported that fossil preparators might resort to visualizing a CT scan of a fossil embedded in a piece of rock to memorize relative bone location as a strategy for technical preparation before they can mechanically remove the surrounding rock.In the absence of tomographic imaging, participants reported the use of textbooks and anatomy diagrams.Figure 3 shows an artifact that P8 uses to form a mental representation of the fossil that is likely to be inside a working piece.According to P5, preparators often see fossils through a microscope at high magnification rates, which requires imagery to promote spatial understanding and situate the current detailed view in relation to the entire object.P1 stated that the preparator work often involves assembling broken bits and pieces to form an anatomically correct fossil ensemble, which certainly requires resorting to the imagery of these specimens.This participant told an anecdote of a laborious Deinotherium assembly that was particularly difficult for being a large specimen scattered into bits and pieces that needed to be imagined together as a jigsaw puzzle, Participants reported the insufficiency of current digital tools in providing paleontology practitioners with appropriate mental representations of objects of interest.P2 stated that during the segmentation selections on "a series of two-dimensional slices, " paleontology researchers struggle to understand what their selections mean for the three-dimensional fossil as they are missing "the big picture of the object".This participant's understanding is that his technical preparator counterparts do not face the same struggle while processing physical objects with their hands.Indeed, technical preparators might have better mental representations than virtual paleontologists.P4, who manages projects involving both kinds of professionals, stated that "unlike scientists, preparators know what is missing from a fossil" in the common scenario of incomplete fossil findings.
Our findings indicate opportunities for the design of novel tools that assist paleontologists in forming more vivid and accurate mental representations of fossils and surrounding materials.Literature indicates the potential of immersion technology, such as virtual reality, to improve spatial understanding and retention [43].Within immersive environments, visuohaptic exploration might potentially improve the performance of tasks that heavily rely on mental imagery.Ernst and Banks describe how humans integrate visual and haptic information in a statistically optimal fashion regarding the reliability of the signals perceived by each modality [10].Incorporating haptic feedback into scientific visualization software could leverage human sensory integration to address some of the challenges that paleontologists face when studying and segmenting low-contrast fossil image datasets.Lacey and Sathian suggest a shared multisensory representation of objects that could underlie visuohaptic cross-modal object recognition [24].In addition to aiding virtual paleontologists in image segmentation tasks, fossil preparators could potentially benefit from rehearsing with a digital tool that allows visuohaptic exploration of specimen-specific scans.

Cognitive Fit
The cognitive fit construct posits that task performance is enhanced when information presentation is compatible with the cognitive processes required to perform a task [50].Our data indicate that virtual paleontologists face challenges regarding the poor cognitive fit between digital fossil preparation and scientific visualization interfaces.P6, P7, and P10 report challenges related to interacting with regions of interest on tomographic slices.The main cognitive fit disconnect occurs because current systems offer two-dimensional interfaces as the main means to interact with three-dimensional data.For example, P10 stated that when he uses the Brush tool in Slicer to create cross-section selections on tomographic slices, checking the selection from another direction reveals many mistakes.P6 described the same issue, adding that performing image segmentation "in one direction, then a second and a third is quite time-consuming".Another task that suffers from cognitive fit is the reorientation of tomographic volumes before segmentation, which P2 described as often necessary as fossils might be scanned from angles that do not yield "characteristic shapes of bones in each of the [tomographic] planes.P10 stated that the process is "not so intuitive" when performed through the use of radial and cubic grids and reorienting the volume.Although scientific visualization software generally offers 3D interfaces for certain operations, such operations still occur through the use of 2D input such as a keyboard and mouse.P2, P3, and P10 also reported using graphic tablets.
Interaction designers wishing to improve the cognitive fit of current scientific visualization interfaces have the opportunity to leverage immersive technologies to match the cognitive requirements of working with 3D data.For example, the implementation of Virtual Reality in this context could empower paleontologists with more intuitive ways to perform tasks related to view manipulation, spatial understanding, and segmentation.The literature describes the benefits of leveraging immersive interfaces for 3D data visualization and interaction [26] as well as for data analysis [23].Amongst our participants, P10 stated that such interfaces could address his aforementioned issues with two-dimensional selections as Virtual Reality would allow him to "start a rough [image] segmentation" with plain conscience of what selections mean "in all three directions, " which "would definitely save a significant amount of time".Literature on existing applications of Virtual Reality for similar freehand interactive tasks, such as 3D sketching, confirm the enhanced cognitive fit of this medium and interaction style as users report increased "perceived appropriateness for the task" as immersion improves "the recognition of spatiality and spatial thinking" [18].Nevertheless, whereas there are evident advantages of employing Virtual Reality within this setting, these benefits are accompanied by corresponding drawbacks that must be taken into consideration.For example, head-mounted displays might offer diminished resolution in comparison with traditional screens that are commonly used in visualization.

Distributed Cognition
Another theme that emerged in our interviews falls under the concept of Distributed Cognition, a human cognition theory that posits that information processing is dispersed across people, their technologies, and their social organizations [16].Amongst fossil preparators, participants P1, P8, and P10 stated that the information needed to complete their preparation tasks resided not only in their existing knowledge but also in the guidance of colleagues.P1 and P8 mentioned seeking the advice of experienced paleontology researchers to obtain information regarding working artifacts.Additionally, P1 stated that certain fossils arrive from excavations completely covered by supporting jackets, in which case only the practitioners who were involved in the initial excavation and collection can provide information regarding the contents of these arrivals.Pieces of information such as fossil site location enable preparators to infer important information such as species, state, and age.Access to such expert advice is, however, not always convenient or timely.As labs specialize in certain types of fossils, experts on exceptional findings might be geographically distant when their expertise becomes necessary.For example, P8 reported traveling to another city to seek advice about a working fossil outside his expertise.
Designers have the opportunity to create interfaces that enable seamless collaboration between technical preparators and paleontology researchers.For example, to address the issue of collaboratively analyzing findings, solutions could leverage the intuitiveness of immersive collaborative environments for discussing fossil data in telepresence.Literature indicates clear advantages of utilizing virtual reality for collaborative data exploration in shared environments [41].For example, in the comparable context of clinical practice, immersive volume rendering provided doctors with "a better spatial understanding for reviewing complex anatomy, " and it excelled at supporting "collaborative discussions [amongst] remote colleagues" [11].Within our study, P4, who manages both technical preparators and paleontology researchers, foresees the potential application of Virtual Reality in enabling researchers to provide preparators with specimen information as part of preliminary preparation rehearsal.

Multimodal Interaction
As we consulted participants regarding the multimodal nature of their workflows, responses varied even within the same role, with differences often attributable to the characteristics of objects of interest.Amongst fossil preparators, P1 reported that her workflow depends solely on vision.Since ammonites are often softer than their surrounding matrix, her workflow involves carefully removing sediment while watching for potential color differences that could possibly signify a fossil surface.P1 explained that she removes sediment at a granular level, constantly watching for flaking to avoid touching fossils with cutting tools as this would cause irreversible destruction.Whereas P1 cannot utilize the sense of touch to guide Ammonite dissection, P8 and P10 described haptics as an essential part of their work.P10 stated depending on the fossil type, "a very soft structure" might indicate that a preparator is "close to a bone," whereas "very hard [structures] in ammonites" might indicate proximity to a shell, all depending "on the preservation of the specimen, age, and structure" as "the older the specimen, the harder the material".According to P8, the majority of his working fossils, arising from oil shale, are harder than their surrounding sediment, which allows him to resort to material compliance as an additional source of information regarding fossil position.While carefully using a set of needles to scrape the oil shale surrounding fossils, P8 attentively listens to the sounds emitted by this interaction to detect "quartz-like sounds" that would indicate material differences and potentially mean that the edge has reached a fossil surface.
Whereas technical preparation might leverage multiple senses, digital workflows are currently restricted to vision.This limitation might impact researchers as participants reported challenges regarding visual perception and its discrimination threshold in regard to low-contrast fossil data.Participants P2, P6, and P7, all of whom have worked with Ammonoids, reported the occasional impossibility of visually discerning between fossils and the sediment matrix surrounding them.According to these participants, Ammonoids often contain dissolved shells due to the effect of physical, chemical, and biological processes over the course of millions of years, which translates into low contrast in tomographic imaging.According to P7, in addition to the discrimination threshold limitations, low-contrast image datasets create difficulties for researchers to fine-tune the settings for threshold-based algorithms to separate fossils from surrounding materials automatically.P2 also reported that he seldom resorts to three-dimensional volume rendering as it lacks visual fidelity and generates occlusion of important features.
As a result of our focus group ideation and subsequent consideration of its outcomes in in-depth interviews, some virtual paleontologists expressed the desire to leverage different senses during digital fossil preparation in a similar fashion to the analog work performed by their technical preparator counterparts.P3 stated that she would welcome a paradigm shift from "working on the [tomographic] planes" and "painting bones" to an interaction style that "would be more like a [physical] preparation" in terms of "[carving] the bones out".P7 favored the "idea of being able to just mill away at certain layers from the outside of the [fossil] system".P6 estimates that haptic feedback "would be a huge advantage" if it could allow him to "feel the shell of [an] ammonite" and "use it for segmentation".P3 estimates that leveraging haptic feedback "would be great to decide between the matrix and the bone" as it occurs "in the real [physical] preparation".Along the same line, P4 estimates that segmentation would be improved if an interface could translate "differences in density" to haptic feedback so that users are "not guessing where boundaries are".P2 stated that he would find an immersive visuohaptic interface to be intuitive "because it is approximately the same that preparators do physically with their hands".Therefore, according to our participant feedback, designers have the opportunity to leverage analog fossil preparation as interaction metaphors that could leverage multimodal interaction and provide users with "instantaneous knowledge about how to interact" with a novel image segmentation interface [17].

Fossil Fragility
Amongst our technical preparator participants, there is a consensus that the fragility of fossil specimens is one of the main challenges impacting their workflows.P1 and P4 stated that fossil fragility is especially concerning when specimens are softer than surrounding materials.For example, P4 reported a particularly challenging preparation that involved removing "shark teeth and fishbone from within limestone".Although preparators process specimens with extreme caution to prevent accidental contact between fossils and cutting tools, bones might be destroyed by the mere proximity to their surfaces.According to P1, certain types of fossils require additional time-consuming prudence as exposing their surfaces demands the immediate application of consolidants to prevent damp materials from falling apart.P10 reported similar concerns over the preparation of "delicate structures [of] very recent bones".
Whereas interactive solutions are unlikely to change the fragile nature of fossil specimens, potential solutions might support the preparation work by preventing destructive contact between tools and specimens.During our focus group discussion, participants involved in fossil preparation expressed their wishes for novel tools that could augment their capacity to predict edges separating sediment and fossil materials.One envisioned solution would be to create a type of sonar that could warn preparators as their tool approaches such edges.Implementing such a system could involve registering and updating a fossil's digital twin, warning practitioners whenever the tracked tool approaches a higher gradient in the tomographic data.Significant changes in material properties could also be performed by a predictive sensing tool such as the artifact described by Picht et al. [36].Selecting a modality to deliver such warnings might be an interesting design problem.For example, audio alerts could be difficult to convey under circumstances such as P1's pneumatic scribe work, which requires wearing hearing protection.Vibrotactile alerts could be a valid alternative medium, although an investigation of human factors would be necessary to provide unobtrusive supplemental sensory cues without taxing technical preparators' haptic attention, as they reportedly rely on haptic cues for material differentiation.
An alternative to edge-detecting warnings would be to provide preparators with a visual overlay that collocates tomographic data and the fossils they represent.As preparators often need to wear eye-protective gear, holographic displays could fulfill this function and provide information regarding the location of fossil features.P4 believes that blending digital and physical fossils would provide preparators the ability to foresee destructive contracts as long as the fossil's digital twin can remain collocated as its physical counterpart changes during preparation.Conveying visual information to preparators might be an interesting design problem.Leveraging visual overlays might be prohibitive in the case of fossil preparators as their workflows require seeing through microscopes or safety glasses, as described by P1 and P8.Even when microscopes are not involved, P1 argued that holographic displays must not reduce visual acuity, given how essential it is for her ability to distinguish materials.

Training
Participants also described the training received by professionals in their fields.According to P1, preparators in Germany often receive training from natural history museums and institutes.Their education includes knowledge of paleontology, anatomy, geology, zoology, and botany.Although these institutions provide students with practical education, it is challenging to train preparators for the specific challenges of their upcoming assignments, as fossil processing widely varies depending on the type of fossil at hand.For example, P1 explained that the preparation of Ammonites is completely different from the work done with oil shale fossils, as the latter requires transfer techniques to address the fragility of such specimens with high water content.P1 and P8 stated that hands-on training of novice preparators ultimately utilizes fossils that are considered disposable.This pragmatic aspect of training creates an inevitable gap in preparation training.
Technological interventions could potentially address the current shortcomings of fossil preparation training.There is potential for technologies such as Virtual Reality and Haptics to emulate the fossil preparation process to train future preparators to process fossils with specific characteristics.P8 stated that the "preparation work itself is experience-driven" and cannot be taught solely by theoretical knowledge.This participant believes in the value of a training system that could be "able to simulate the feeling of sweeping a needle on the scale of the fish or on a bone".P9 would find it valuable to train preparators through digital tools as a means of educating students without incurring the risk of damaging fossils.Evidently, creating a simulation that could accurately emulate parts of the preparation work would be challenging from a technical standpoint.Haptics would be an essential component of any fossil preparation simulation.P9 stated that it is "important to know how the stone feels".P8 and P9 demonstrated some skepticism about the ability of a computer system to accurately emulate a process as complex as the interaction with different fossil materials.P9 emphasized the challenges of making such a simulation realistic, such as accurately representing the subsequent effects of touching a fossil in certain areas.

DISCUSSION
This study aimed to investigate the challenges encountered by technical preparators and paleontology researchers in analog and digital fossil preparation.We employed qualitative methods to gather the issues these professionals face in regard to their tools and methods, and we provided human-computer interaction practitioners with a detailed report of addressable fossil preparation issues along with potential opportunities for design interventions that could improve current paleontological workflows.
Our findings mainly differ from existing literature due to our focus on the interactive aspects of preparation tools and methods.Previous works describing analog and digital methods are mainly targeted toward scientists, informing them about both established and experimental tools and methods.For example, in 2014, Cunningham et al. introduced the scientific community to Virtual Paleontology methods that were revolutionizing the field [8].Regarding traditional fossil preparation, Leiggi et al. offer a comprehensive guide of tools and methods involved in "specimen collection, preparation, conservation, and reproduction" [28].Other works aimed to inform members of the general public of interesting details of the fascinating work performed in preparation labs [53].Therefore, the present study fills a previous gap in investigating this fossil preparation through the human-computer interaction lens.
Although our study had a different focus from the existing literature, our findings are in line with previous work in terms of reported workflow challenges.For example, participant statements regarding the issues caused by low attenuation contrast yielded by certain fossils align with the problems reported by Sutton et al. [47].The same can be said about the hardships imposed by the fragility of certain specimens [4].Our results differ from the literature when it comes to some of the unique techniques described by participants, which reflect the creative aspect of the preparation craftsmanship and the resourcefulness of preparators in addressing issues that guiding texts cannot entirely predict.
Gathering technical preparators and virtual paleontologists into a single focus group session proved to be fruitful as it enabled productive exchanges and ultimately ideated potential solutions for their counterparts' workflows.The contextual inquiries that ensued provided us with the possibility to obtain expert feedback regarding solutions that were ideated during our focus group.This combination of qualitative methods allowed us to validate ideas and enabled us to present interaction researchers with important design considerations regarding preparation tools.It is important to note that our findings should be read in the context of this methodological approach to consider any potential biases that this sequence of events might have introduced.
One important limitation of our study is its number of participants.As the present study reached ten practitioners for relatively short inquiries, our investigation could not match the breadth and depth of larger ethnographic studies such as the work conducted by Wylie [54].This modest participant count also implicates a constraint in geographical terms, as we engaged participants from only two countries.Another inference of our limited sample size regards the specific types of fossils and respective preparation tools and methods that could be elicited by our participants.Given the implications of our number of participants, our investigation may face validity threats.For example, our findings may have limited external validity as they might not be transferable to the fossil preparation work of fossils other than ammonites and oil shale mammals.In order to allow readers to reach informed conclusions about the transferability of our findings, we provide detailed descriptions of our participants' contexts within the reasonable limits for this report's format and participant privacy.

CONCLUSION
In the present study, we leveraged qualitative methods to elicit the challenges that traditional and modern fossil preparators face regarding their tools and methods.We analyzed our data through a human-computer interaction perspective and unveiled opportunities for the design of novel artifacts that address the unique challenges of analog and digital fossil preparation.
Based on the scope limitations of this study, future research might complement our effort by learning from fossil preparators who process specimen categories and respective preparation methods that this investigation could not comprise.Another research direction could be the validation and potential implementation of the presented design opportunities through user-centered design methods.
Fossil preparation is a relatively small field that yields limited commercial incentives for industries to create novel tools to improve such paleontological workflows.Therefore, we believe that the main implication of our contribution is to inform HCI researchers of fossil preparation challenges, as we conclude that only researchers will have the ability and willingness to tackle these design opportunities and support fossil preparators.

Figure 1 :
Figure 1: A technical preparator (P1) alternates between using a fine-point carving tool and an air scribe to perform Matrix Removal.

Figure 2 :
Figure 2: A Screen Capture Depicting a Low-Contrast Trabecular Bone Slice in ImageJ: P2 highlighted this image as a prime example of tomographic data requiring manual processing.

Figure 3 :
Figure 3: A collection of fossil anatomy diagrams that a technical preparator (P8) consults to form mental representations of working fossils.