Sketchy understandings: Drawings reveal where students may need additional support to understand scale and abstraction in common representations of DNA

8 Visual representations in molecular biology tend to follow a set of shared conventions for using certain 9 shapes and symbols to convey information about the size and structure of nucleotides, genes, and 10 chromosomes. Understanding how and why biologists use these conventions to represent DNA is a key 11 part of visual literacy in molecular biology. Visual literacy, which is the ability to read and interpret 12 visual representations, encompasses a set of skills that are necessary for biologists to effectively use 13 models to communicate about molecular structures that cannot be directly observed. To gauge students’ 14 visual literacy skills, we conducted semi-structured interviews with undergraduate students who had 15 completed at least a year of biology courses. We asked students to draw and interpret figures of 16 nucleotides, genes, and chromosomes, and we analyzed their drawings for adherence to conventions for 17 representing scale and abstraction. We found that 77% of students made errors in representing scale and 18 86% of students made errors in representing abstraction. We also observed about half of the students in 19 our sample using the conventional shapes and symbols to represent DNA in unconventional ways. These 20 unconventional sketches may signal an incomplete understanding of the structure and function of DNA. 21 Our findings indicate that students may need additional instructional support to interpret the conventions 22 in common representations of DNA. We highlight opportunities for instructors to scaffold visual literacy 23 skills into their teaching to help students better understand visual conventions for representing scale and 24 abstraction in molecular biology. 25 26 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 3

27 Biologists rely on visual models to help them communicate about complex molecular structures 28 that cannot be seen with the human eye. Such visual models tend to convey biology concepts through 29 established discipline-specific visual conventions (1), and there are particular sets of shapes and symbols 30 that biologists use to reduce complexity and to emphasize the most salient features of molecular structures 31 for a given context (2). Consider how biologists often use a ladder shape to simply model a short DNA 32 sequence. Conventionally, the sides of the ladder represent a sugar-phosphate backbone and the rungs of 33 the ladder represent the nucleobase pairs. A biologist may also conventionally communicate about short 34 DNA sequences using letters or space-filling models when the specific order of nucleotides or the 35 chemical structure of DNA is the most relevant feature they want to emphasize. Understanding how and 36 when to use such conventions to convey meaning is a key part of visual literacy in molecular biology (3, 37 4). 38 Like its counterpart in written communication, visual literacy in molecular biology refers to the 39 ability to “read” and “write” using the conventions of discipline-specific symbols and notations that 40 encode meaning in visual representations (1, 5). Experts in molecular biology visual literacy can 41 accurately “read” or decode how conventions are used in a visual model to convey pertinent information 42 and can “write” or draw a representation using conventions to communicate biology concepts to others 43 (6). Developing expertise in molecular biology visual literacy is an integral part of molecular biology 44 education. Visual literacy is so central to molecular biology education that the skills associated with 45 “reading” and “writing” using visual models are encompassed within the Vision and Change core 46 competencies for “Modeling” and “Communication and Collaboration” (7, 8). 47 Despite the importance of visual literacy in molecular biology, our previous work indicated that 48 undergraduate students may need additional support to develop expert-level visual literacy for molecular 49 biology concepts (9, 10). Developing visual literacy may be particularly challenging for students because 50 biologists can represent molecular structures at different scales and often use different abstract symbols to 51 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 4 emphasize varying aspects of the same structure (1, 5, 11, 12). Previous research hypothesized that 52 challenges in understanding the conventions of scale and abstraction in visual models may be primary 53 constraints on the development of visual literacy (6, 13). 54 The DNA Landscape (12) is a useful conceptual framework for delineating the conventions of 55 how biologists represent DNA at different levels of scale and abstraction. The DNA Landscape was 56 developed after reviewing how DNA was represented across thousands of textbook figures, and the 57 resulting conceptual framework is presented as a three-by-three matrix. Representations of DNA can be 58 mapped to a location on the DNA Landscape based on scale (nucleotide, gene, chromosome) and the 59 degree of abstraction (literal shape, elements of shape and abstraction, very abstract). The DNA 60 Landscape emerged from an analysis of textbooks that were written and edited by biology experts, so the 61 way that scale and abstraction is portrayed in the DNA Landscape reflects the conventions accepted and 62 used by biology experts. 63 We can use the DNA Landscape framework to identify erroneous or unconventional ways that 64 biology students are “writing” with the conventions that biologists use to represent DNA. If a student’s 65 sketch (or the verbal description of what their sketch represents) does not follow the conventions typically 66 associated with a specific location on the DNA Landscape, this signals that the student may not correctly 67 understand how and why biologists use certain symbols to represent DNA. We anticipate that misuse or 68 misunderstanding of the typical visual conventions for representing DNA is an indicator that the student 69 likely holds deeper conceptual errors about DNA structure or function. 70 Proficiency in molecular biology visual literacy relies on a foundational content knowledge in 71 biology as well as an understanding of how biologists conventionally represent such content in diagrams 72 and figures. Here, we designed an interview protocol to specifically probe the content knowledge and 73 visual literacy skills of undergraduate students, in which students were asked to sketch representations of 74 DNA. We analyzed student sketches generated during the interview protocol to answer the following 75 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 5 research question: “In what ways are students misusing or misunderstanding the common conventions for 76 representing scale and abstraction in molecular biology?” 77

78 We conducted 45-minute semi-structured interviews with 35 students from two institutions 79 (Table 1). All research participants had completed at least a year of undergraduate biology coursework. 80 We recruited participants via institutional email, flyers, in-person recruitment during upper-level biology 81 courses, and snowball sampling. During the recruitment process, we notified participants that they would 82 be completing sketches during the interview and we asked them to arrive to the interview prepared with a 83 pen and paper. We compensated participants with a $20 Amazon gift card. This research was approved 84 and classified as exempt from human-subjects review by Rochester Institute of Technology (protocol 85 01090823). 86 Our interview protocol involved a series of tasks designed to elicit specific molecular biology 87 visual literacy skills across levels of Bloom’s Taxonomy (14, 15). The protocol was composed of three 88 parts which followed the same series of tasks that were modified in each part to focus on a different topic 89 in molecular biology (chromosomes, nucleotides, genes). We present a generalized and a specific set of 90 interview tasks in Table 2. 91 Within the interview protocol, we asked participants to explain figures from published concept 92 assessment instruments when the figures were presented in isolation, and then asked them to answer the 93 entire concept assessment item in a later task during the interview. The concept assessment items were 94 Item #2 from the Genetics Concept Assessment (16), Item #10067 from the BioMolViz assessment 95 library (17), and Item #38 from GenBio-MAPS (18), which included an abstract model of unreplicated 96 chromosomes, a space-filling model of DNA nucleotides, and a box-and-line representation of a gene, 97 respectively. We used these items because they each included a conventional representation of a 98 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 6 molecular biology structure and the items had been vetted through rigorous assessment validation 99 procedures by the developers of each assessment. 100 We conducted and recorded interviews using Zoom. Of the 35 interviews, 11 participants created 101 drawings using the Zoom whiteboard and 24 participants drew on paper then subsequently sent us photos 102 or scans of their sketches via email. The quality of the drawings and sketches on the Zoom whiteboards 103 may have varied depending on the familiarity and comfort with the platform. 104 Students generated 277 unique sketches of nucleotides, genes, and chromosomes during the 105 interviews. We coded each sketch for alignment to the DNA Landscape (12). Using the visual 106 conventions within the DNA Landscape as a reference, we used structural coding (19) to code each sketch 107 for erroneous or unconventional uses of scale and abstraction (Table 3). We considered sketches in the 108 context of how students described what they drew. In some cases, participants drew what appeared to be 109 an unconventional representation but verbally acknowledged and clarified the ways in which their sketch 110 differed from their mental image of what they were trying to represent. When the participant followed 111 their sketch with a verbal description of how they would amend their sketch to better align with typical 112 conventions, we considered these sketches to be conventional. 113 To establish interrater reliability, we randomly selected 10% of the sketches in the sample (n = 114 27) and two raters (CU and DLN) independently reviewed the sketches and associated transcript for 115 alignment to the DNA Landscape and for the use of erroneous or unconventional scale and abstraction. 116 The two raters met to discuss any disagreements. We refined our codebook based on where disagreements 117 occurred. The two raters repeated this process with a second set of sketches (n = 27). After establishing 118 interrater reliability in independent coding, one rater coded the remainder of sketches in the sample. Both 119 raters subsequently met to review the application of codes to the entire sample, and consensus values for 120 each sketch are reflected in the final dataset. 121 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 7

122 We found that the majority of students think about the commonly-used shapes and symbols in 123 molecular biology representations in different ways than experts might. Across the 35 students we 124 interviewed, 97% (n = 34) drew at least one sketch that represented nucleotides, genes, or chromosomes, 125 in ways that were misaligned to the conventional ways that experts typically represent the same 126 structures. Here, we identify the types of molecular structures students drew and the frequency of the 127 misuses or misunderstandings of the common conventions for representing scale and abstraction in 128 molecular biology. We provide exemplar sketches that illustrate the ways that students were using shapes 129 and symbols in ways that suggest an incomplete understanding of foundational molecular biology 130 concepts. 131 What did students draw? 132 We identified 277 student sketches that aligned to locations within the DNA Landscape (Figure 133 1). The majority of sketches were “highly abstract” representations consisting of letters or simple shapes. 134 Sketches often contained multiple elements from across the DNA Landscape. These findings are not 135 necessarily surprising, as experts often communicate with abstract representations across scales. The 136 higher incidences of chromosome representations may reflect that chromosomes were the first subject in 137 the sequence of the interview protocol and that many students drew chromosome-scale structures in their 138 depictions of genes (e.g., representing a gene as a chunk of a chromosome arm). 139 How often did sketches contain errors or unconventional representations of scale 140 and abstraction? 141 Overall, approximately half of the sketches in our sample (49%, n = 137) reflected erroneous or 142 unconventional depictions of scale and/or abstraction to represent structures in molecular biology (Table 143 4). While only about a third of sketches contained errors in abstraction (29%), we found that 86% of 144 students drew at least one sketch with an abstraction error. We similarly found that even though the total 145 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 8 number of sketches with scale errors was relatively small (17% of sketches), over three-quarters of 146 students made a scale error in at least one sketch. Unconventional abstraction was less common, 147 comprising only 10% of the sketches. Notably, we did not identify any instances of unconventional scale. 148 We found that there were associations between the location of a sketch on the DNA Landscape 149 and the frequency of errors in scale or abstraction. The “highly abstract” representations in the the DNA 150 Landscape (nucleotide sequence, gene name, Chromosome X) often had higher frequencies of abstraction 151 errors (Figure 2). While abstraction errors occurred across nearly all locations in the DNA Landscape, 152 there was a clear pattern that conventions of “highly abstract” representations were a frequent source of 153 misunderstanding for students. We also identified student sketches with scale errors across many of the 154 DNA Landscape locations, with the most frequent scale errors occurring in sketches of nucleotide ladders 155 and gene helices. 156 What did errors in scale look like? 157 We provide a few illustrative examples of errors in representing scale in Figure 3. Many of the 158 errors in scale reflected a misunderstanding the conventions for representing the size of genes. Consider 159 how both Alex and Donna used the helical and ladder representations of DNA, respectively, to indicate 160 that the length of a gene is approximately 3 nucleotides. We see additional misuses of conventional gene 161 symbolism in Alex’s sketch in which the start and stop codons are boxes approximately the same size as 162 the boxes for introns and exons in a typical box-and-line representation. We want to emphasize the 163 mismatch between the sketches that Alex and Donna drew and the words that they used to describe their 164 sketches. Alex demonstrates an understanding that a gene is a “section of the DNA that codes for your 165 proteins,” and Donna similarly states that “a gene is a sequence of DNA,” yet the sketches each student 166 drew indicate that they are substantially underestimating just how “big” the section or sequence of DNA 167 is in a typical gene. While Alex and Donna underestimated the size of a gene, we saw Sheryl making 168 overestimates. Sheryl drew a single gene, labeled as SRY in her sketch, which comprises nearly half the 169 length of an arm of a chromosome. All the examples in Figure 3 notably contain elements from multiple 170 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 9 locations on the DNA Landscape, which was a common, but not defining, characteristic of many of the 171 sketches that contained errors in scale. 172 What did errors in abstraction look like? 173 Many of the errors in abstraction occurred in the highly-abstract representations of chromosomes 174 and genes. We have an extensive discussion of the common errors in drawing and interpreting highly-175

X-shaped chromosomes and Punnett squares (which are included in the highly-abstract gene 176 name location on the DNA Landscape) in our previous research (10), so we focus on abstraction errors in 177 other locations on the DNA Landscape in Figure 4. 178 We were surprised to see how many different ways students were misinterpreting the 179 conventional double-helix shape of DNA, and we include two examples here. Brittany interpreted the 180 phrase “double helix” to mean that there were two strands on each side of the structure. When completing 181 her sketch, she used two pen strokes on each side of the twisted X-shape to emphasize the “double” 182 nature of the double helix. Brittany’s drawing also indicated a misunderstanding of the phrase “backbone” 183 that is conventionally used to refer to the alternating deoxyribose sugars and phosphate groups in a DNA 184 strand. In her sketch, Brittany referred to the horizontal rung-shaped lines as the backbone of DNA and 185 described this horizontal backbone as where DNA “is held.” She wrote out a series of three letters 186 typically associated with nucleobases that were “held” upon the horizontal lines as if upon a shelf. 187 In comparison, Norma’s sketch of DNA is much more conventional in appearance. At first 188 glance, the double-helix shape looks like what an expert might quickly draw to illustrate the double-helix 189 shape of DNA. However, when Norma described her sketch, we can see that she has misunderstood the 190 conventions — her double- stranded helix was actually representing “two RNA strands.” While Norma 191 replicated a conventional drawing of DNA, she had an unconventional understanding of how experts 192 typically use the double-helix shape to convey meaning about the two-stranded structure of DNA. 193 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 10 We also found several errors in the way students think about the meaning of conventional box-194 and-line diagrams of genes. Lucy’s gene here is entirely composed of boxes — the enhancer, promoter, 195 exons, and poly A tail in the sketch are all represented with boxes. The line in Lucy’s box-and-line is only 196 represented here after the end of the sequence of boxes has ended, which contrasts with the more 197 conventional use of lines to illustrate untranscribed regions in genes. Lucy also mixes conventions of 198 representing DNA and mRNA. Note how the “coding regions” of the gene consisting exclusively of 199 exons, as if the gene had already undergone splicing. The gene in this sketch also resembles mRNA in its 200 inclusion of a poly A tail. Despite multiple misuses of the conventions, Lucy insisted that this type of 201 diagram was what she had seen in “all of [her] classes.” 202 What does unconventional abstraction look like? 203 Student sketches were unconventional when the way that a student depicted a nucleotide, gene, or 204 chromosome was not necessarily incorrect but did not match the standard uses of shapes, symbols, or 205 sometimes colors, that experts typically rely on to convey meaning about molecular biology topics. In 206 Figure 5, we see that students often use the correct vocabulary to describe molecular structures yet their 207 sketches leave remaining questions about whether they truly understand the vocabulary they used. 208 Consider the two chromatids in Sylvia’s sketch of a chromosome. Sylvia correctly states that 209 chromosomes are made of two chromatids, yet she only colored in one of the chromatids in the pair. This 210 color coding is unconventional because experts often use different colors to visually distinguish between 211 non-homologous chromosomes in diagrams of mitosis or in karyotypes. Sister chromatids contain 212 genetically identical information, so coloring only one of the chromatids conventionally signals a 213 difference between the two. While Sylvia uses the correct words, we do not know if she robustly 214 understands chromosome structure based on her drawing. 215 We see a similar situation where correct words are paired with unclear symbolism in Socks’ box-216 and-line sketch of a gene. Here, the boxes are not conventionally representing exons, but instead are 217 supposed to be operons, promoters, and TATA boxes. This sketch also uses the narrow line to represent 218 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 11 both introns and exons, distinguishing the two by unconventional color coding. While someone with more 219 expertise might use similar words and similar shapes to describe a gene, the labels and color coding that 220 Socks uses to explain the sketch leave questions about the extent to which they understand gene structure. 221 In a third case, Evelyn represents a gene as a narrow band on a chromosome, which could be a 222 fairly conventional representation of genes. However, the chromosome is replicated, and the gene is only 223 indicated on one of two arms. We do not know from this sketch or from her description whether Evelyn 224 understands that the same gene would also be present in the same location on other sister chromatid. 225 What does it look like when sketches contain both errors in scale and abstraction? 226 While only a small portion of sketches contained both errors in scale and abstraction (n = 14), the 227 types of errors we observed in these sketches indicate deep misunderstandings of foundational structures 228 in molecular biology. These sketches incorrectly mixed and matched the conventions across locations on 229 the DNA Landscape. We describe three such instances of students misusing conventions for scale and 230 abstraction in Figure 6. 231 Similar to Donna (Figure 3), Heather misused the conventions of the nucleotide ladder to 232 represent exons and introns as structures consisting of approximately 3 nucleotide base pairs. Heather 233 drew this sketch as a representation of the true-false statement in GenBio-MAPS: “Only mutations that 234 occur in the exons can have an impact on the cell,” and chose to represent the adenine-adenine pairing as 235 a mutation that might occur. Heather correctly transcribed DNA into mRNA, but continued to use the 236 conventions of the ladder to represent mRNA as a double-stranded structure. 237 Christin’s sketch of a gene indicated a misunderstanding of both the conventions of abstract X-238 shaped chromosomes as well as the scale of a gene. In the first part of the interview, Christin originally 239 drew a chromosome as two side-by-side lines. When we asked Christin to draw a gene later in the 240 interview, she drew a gene using the same shape, size, and length as just one of the lines in the original 241 chromosome sketch. She verbally indicated that her sketch showed that a single gene was the entire 242 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 12 length of a chromosome. While both genes and chromosomes can be represented with lines, Christin 243 misunderstood that the similarly-shaped lines representing genes and chromosomes conventionally 244 indicate vastly different scales of genetic information. 245 Annie was among many of the participants who substantially misrepresented the scale of 246 nucleotides and proteins. Here, Annie represents a protein as a smiling hexagon that is binding to the –OH 247 group of a single DNA nucleotide. In addition to representing an entire protein at the same scale as a 248 single nucleotide, Annie makes an interesting choice to use a hexagon to represent the protein structure. 249 While proteins can be abstractly represented using a variety of shapes, ranging from circles to amorphous 250 blobs, experts rarely use hexagons to represent proteins because hexagons are conventionally used to 251 represent sugars. 252 What are students doing to accurately convey scale? 253 We were encouraged to see that some students were cognizant of the limitations of accurately conveying 254 scale in a quick sketch. Several students acknowledged how they would like to improve their sketch to be 255 more realistic. For example, Piper drew a short sequence of letters to represent a gene and used ellipsis 256 and the label “# of bps [base pairs] long” to signal that genes are much longer than what is feasible to 257 represent in a sketch (Figure 7). While students like Piper demonstrated their understanding of scale by 258 pointing out the shortfalls in their sketches, this was not common across all students. Unless the student 259 verbally or visually indicated a limitation of their sketch, we assumed that students’ sketches were 260 representative of their mental models of scale. 261

262 We analyzed student sketches of nucleotides, genes, and chromosomes and found that most 263 students misused or misunderstood the visual conventions for scale and abstraction that experts typically 264 use to represent these molecular structures. Some of the most recognizable symbols in biology — the 265 double helix and the X-shaped chromosome — were among the most commonly misunderstood. 266 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 13 Understanding scale and abstraction in molecular structures is foundational for visual literacy (6, 13), and 267 our findings suggest that undergraduate students may need additional instructional supports to develop 268 fluent visual literacy in molecular biology. 269 Supporting student understanding of conventions for scale 270 Scale is so important to science that it is included as a crosscutting concept in a widely-used 271 framework for science education (20), yet scale is rarely emphasized or assessed in undergraduate biology 272 courses (21). To help better incorporate scale into biology teaching and learning, we recommend using 273 appropriately scaled physical models of molecular structures to provide a foundation for student thinking 274 (22). When using physical models in a classroom may be limited by monetary or class size constraints, 275 making analogies to familiar objects can also be a fruitful way to build connections in student 276 understanding of molecular scale (23). For example, when discussing the size of a gene or the size of a 277 chromosome, instructors can make a comparison to a skein of yarn. Instructors can challenge students to 278 determine how many inches a gene would be if nucleotides are scaled to the size of a single strand of 279 yarn, and to consider the relationship between the wrapped nature of a skein and the condensed nature of 280 a chromosome. Relating nucleotides, genes, and chromosomes to a physical object like yarn may help 281 students visualize that a gene can be tens of thousands of nucleotides, and might help navigate students 282 away from misunderstandings like a single gene comprises a substantial portion (or the entire length) of a 283 chromosome. After using yarn as an example, we suggest instructors ask students to translate that analogy 284 back to the typical conventions for representing scale in molecular biology. Instructors can show various 285 examples of sample sketches in which genes are represented at conventional and unconventional sizes and 286 ask students to discuss in groups and explain which representation is most accurately portrays the scale of 287 a gene. 288 Instructors may also consider asking their class to evaluate each other’s drawings of scale. In this 289 study, we asked students to “zoom in” and “zoom out” of molecular structures, and these prompts 290 generated some of the most interesting misrepresentations of scale in student sketches. Instructors can 291 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 14 provide students with drawing materials in class, ask students to “zoom in” and “zoom out” from 292 nucleotides, genes, and/or chromosomes, and then ask students to evaluate each other’s sketches. In such 293 an approach, we suggest instructors end the activity with their own expert analysis of student sketches to 294 positively reinforce which conventions were used correctly and steer students away from incorrect 295 representations of scale. 296 Supporting student understanding of conventions in abstract representations of 297 DNA 298 Biologists have multiple ways of visually representing the same molecular structures, but these 299 multiple representations can pose challenges for students. While experts can fluently translate between 300 representations of nucleotides as letters, as ladders, and as chemical structures because their experience 301 using such symbols has conferred “representational competence” (24), students may need additional 302 guidance to correctly choose which representation is most appropriate to communicate a certain point. To 303 help students develop this representational competence, instructors can be explicit about how biologists 304 use abstraction to focus viewers on the most salient features of the structure. Instructors may ask students 305 to consider why a biologist might use letters rather than chemical structures when the most important 306 information to convey to another biologist is the sequence of nucleobases. To help students understand 307 why there are multiple levels of abstraction for representing the same molecular structures, we 308 recommend using active-learning opportunities in class where students discuss in small groups why and 309 when a biologist might use certain representations to communicate. Instructors may also consider 310 incorporating “figure analysis” activities in which students work in groups to examine and discuss the 311 meaning and the limitations of the abstract symbolism in textbook figures (25). 312 In addition to discussing polished textbook figures, instructors may find it a useful and insightful 313 learning task to have students evaluate unconventional representations of DNA. Instructors may draw 314 their own intentionally unconventional sketches, find examples of unconventional representations 315 published online in image banks such as Adobe Stock, or create unconventional representations using 316 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 15 artificial intelligence. Asking students to consider the strengths and limitations of a representation and 317 how the representation conventionally or unconventionally portrays molecular structures may help 318 students understand what features in abstractions enable effective communication between biologists. 319 Using the DNA Landscape to support student understanding of scale and 320 abstraction 321 Our findings suggest that most students may need support to understand and use visual 322 conventions for representing DNA. We recommend instructors use the DNA Landscape (12) as an 323 instructional tool to help scaffold their teaching about visual representations of DNA. When designing a 324 lesson, a unit, or a course, instructors should consider how they are incorporating representations of DNA 325 from locations on the DNA Landscape. Instructors can explicitly compare and contrast representations 326 within the same column, highlighting how similar meaning about a structure is conveyed at different 327 levels of abstraction. We also recommend that instructors consult the DNA Landscape when they are 328 using representations that span across columns, as spanning columns within a representation can muddle 329 scalar relationships. For example, we saw that many students were representing entire genes (from the 330 middle column of the DNA Landscape) with letters (from the left column of the DNA Landscape), and 331 they typically represented entire genes with fewer than 10 letters. Mixing and matching across columns 332 can create confusion about scale. If using letters to represent a gene, instructors may consider following 333 Piper’s example in Figure 7 and explicitly writing out an estimate for the number of letters that might 334 compose the entire gene. 335 If instructors emphasize visual literacy skills in their teaching, they should make sure that these 336 priorities are also reflected in their assessments. Despite the frequency with which visual models are used 337 in teaching, we previously found that visual models are largely absent from undergraduate biology exams 338 (26). If instructors are teaching students using representations from across the DNA Landscape, we 339 encourage instructors to also assess students on their understanding of representations from across the 340 DNA Landscape. In this present study, we found that drawings were an effective way to quickly assess 341 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 16 student understanding of molecular structures, but we were only able to glean many of these 342 understandings by comparing both sketches and verbal descriptions (such as in Norma’s double-stranded 343 helix representing two RNA strands in Figure 4). We recommend that instructors consult the Three-344 Dimensional Learning Assessment Protocol (27) for criteria that may guide the creation of assessment 345 questions that can engage students in productively reasoning about visual models. 346

347 We necessarily focused our analysis on student errors and misunderstandings to sharpen our 348 focus on the areas of undergraduate biology instruction that may need additional scaffolding and support. 349 This decision was not meant to undervalue students’ productive thinking about molecular biology. 350 Our sample consisted of students from two institutions, and such, the types of errors in scale and 351 abstraction that we observed may not be representative of student thinking in other populations. 352

353 Biologists rely on conventional visual models to communicate about molecular structures, yet the 354 conventions in such models are often misunderstood by undergraduate biology students. Even though the 355 shapes and symbols to represent nucleotides, genes, and chromosomes are ubiquitous in undergraduate 356 biology courses, students may not be interpreting the meaning of the shapes and symbols in the same 357 ways as experts. By asking students to draw molecular structures, we saw that many students had sketchy 358 understandings of foundational biology concepts. As visual literacy is central for communicating about 359 molecular biology, we recommend instructors closely examine the visual representations they use in their 360 courses and consider the ways in which they can reinforce student understanding of the conventions for 361 representing scale and abstraction. 362

363 We thank the students who participated in this research. We thank Kerstyn Gay for creating the 364 illustrations in the DNA Landscape in Figure 1. Funding provided by NSF DGE 2222337. Any opinions, 365 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 17 findings, and conclusions or recommendations expressed in this material are those of the author(s) and do 366 not necessarily reflect the views of the National Science Foundation. 367 368 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 18

369 1. Schönborn KJ, Anderson TR. 2006. The importance of visual literacy in the education of 370 biochemists. Biochem Mol Biol Educ 34:94–102. 371 2. Quillin K, Thomas S. 2015. Drawing- to-Learn: A framework for using drawings to promote model-372 based reasoning in biology. CBE—Life Sci Educ 14:es2. 373 3. Messaris P, Moriarty S. 2004. Visual Literacy Theory, p. 479–502. In Smith, K, Moriarty, S, 374 Barbatsis, G, Kenney, K (eds.), Handbook of Visual Communication. Routledge, New York, NY. 375 4. Trumbo J. 1999. Visual literacy and science communication. Sci Commun 20:409–425. 376 5. Offerdahl EG, Arneson JB, Byrne N. 2017. Lighten th e load: Scaffolding visual literacy in 377 biochemistry and molecular biology. CBE – Life Sci Educ 16:es1. 378 6. Schönborn KJ, Anderson TR. 2009. A model of factors determining students’ ability to interpret 379 external representations in biochemistry. Int J Sci Educ 31:193–232. 380 7. AAAS. 2011. Vision and Change in Undergraduate Biology Education: A Call to Action. Final 381 Report of a National Conference Organized by the AAAS with support from NSF. American 382 Association for the Advancement of Science, Washington, D.C. 383 8. Clemmons AW, Timbrook J, Herron JC, Crowe AJ. 2020. BioSkills Guide: Development and 384 national validation of a tool for interpreting the Vision and Change core competencies. 4. CBE—385 Life Sci Educ 19:ar53. 386 9. Newman DL, Spector H, Neuenschwander A, Mill er AJ, Trumpore L, Wright LK. 2023. Visual 387 literacy of molecular biology revealed through a card-sorting task. 0. J Microbiol Biol Educ 388 24:e00198-22. 389 10. Uminski C, Newman DL, Wright LK. 2025. Probing visual literacy skills reveals unexpected 390 student conceptions of chromosomes. CBE—Life Sci Educ 24:ar17. 391 11. Kindfield ACH. 1994. Biology diagrams: Tools to think with. J Learn Sci 3:1–36. 392 12. Wright LK, Wrightstone E, Trumpore L, Steele J, Abid DM, Newman DL. 2022. The DNA 393 Landscape: Development and application of a new framework for visual communication about 394 DNA. CBE—Life Sci Educ 21:ar47. 395 13. Schönborn KJ, Anderson TR. 2010. Bridging the educational research- teaching practice gap. 396 Biochem Mol Biol Educ 38:347–354. 397 14. Anderson LW, Krathwohl DR, Bloom BS. 2001. A Taxonomy for Learning, Teaching, and 398 Assessing: A Revision of Bloom’s Taxonomy of Educational Objectives. Longman, New York, 399 NY. 400 15. Arneson JB, Offerdahl EG. 2018. Visual literacy in Bloom: Using Bloom’s Taxonomy to support 401 visual learning skills. CBE — Life Sci Educ 17:ar7. 402 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 19 16. Smith MK, Wood WB, Knight JK. 2008. The Genetics Concept Assessment: A new concept 403 inventory for gauging student understanding of genetics. CBE—Life Sci Educ 7:422–430. 404 17. Beckham J, Acevedo R, Dries DR, Engelman S, Franzen MA, Jakubowski HV, Mertz P, Novak 405 WRP, Roberts RA, Roca AI, Procko K. 2023. The BioMolViz Library: An Assessment Repository 406 to Evaluate Biomolecular Visual Literacy Through the BioMolViz Framework. University of Texas 407 at Austin. 408 18. Couch BA, Wright CD, Freeman S, Knight JK, Semsar K, Smith MK, Summers MM, Zheng Y, 409 Crowe AJ, Brownell SE. 2019. GenBio-MAPS: A programmatic assessment to measure student 410 understanding of Vision and Change core concepts across general biology programs. 1. CBE—Life 411 Sci Educ 18:ar1. 412 19. Saldaña J. 2013. The coding manual for qualitative researchers, 2nd ed. SAGE, Los Angeles. 413 20. National Research Council. 2012. A Framework for K-12 Science Education: Practices, 414 Crosscutting Concepts, and Core Ideas. National Academies Press, Washington, D.C. 415 https://www.nap.edu/catalog/13165/a-framework-for-k-12-science-education-practices-416 crosscutting-concepts. Retrieved 18 July 2021. 417 21. Uminski C, Burbach SM, Couch BA. 2024. Undergraduate biology lecture courses predominantly 418 test facts about science rather than scientific practices. CBE—Life Sci Educ 23:ar19. 419 22. Newman DL, Stefkovich M, Clasen C, Franzen MA, Wright LK. 2018. Physical models can 420 provide superior learning opportunities beyond the benefits of active engagements. Biochem Mol 421 Biol Educ 46:435–444. 422 23. Newman DL, Uminski C, Wright LK. 2024. Student- generated analogies for learning about 423 information flow. CourseSource 11. 424 24. Kozma RB, Russell J. 1997. Multimedia and understanding: Expert and novice responses to 425 different representations of chemical phenomena. J Res Sci Teach 34:949–968. 426 25. Wiles AM. 2016. Figure analysis: A teaching technique to promote visual literacy and active 427 Learning. Biochem Mol Biol Educ 44:336–344. 428 26. Uminski C, Cammarota C, Couch BA, Wright LK, Newman DL. 2024. Biology exams rarely use 429 visual models to engage higher-order cognitive skills. bioRxiv 430 https://doi.org/10.1101/2024.12.23.630136. 431 27. Laverty JT, Underwood SM, Matz RL, Posey LA, Carmel JH, Caballero MD, Fata- Hartley CL, 432 Ebert-May D, Jardeleza SE, Cooper MM. 2016. Characterizing college science assessments: The 433 Three-Dimensional Learning Assessment Protocol. 9. PLOS ONE 11:e0162333. 434 435 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 20 436 437 Figure 1. Number of student sketches aligned to each location on the DNA Landscape. Sketches that 438 contained elements from multiple locations on the DNA Landscape were coded in each location, so the 439 sum of sketches here exceeds the 277 total sketches generated in the interviews. 440 441 442 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 21 443 444 Figure 2. Heat map indicating the frequency of scale errors, abstraction errors, or unconventional 445 abstraction in sketches aligned to locations on the DNA Landscape. Sketches with errors in scale 446 reflect a factual error in their understanding of the size of a particular molecular structure or the scalar 447 relationship between multiple molecular structures. Sketches with errors in abstraction reflect a 448 misunderstanding of the visual conventions that are typically used to represent a particular molecular 449 structure or concept. Sketches with unconventional abstraction convey factually correct information about 450 the size of a particular molecular structure or the scalar relationship between multiple molecular 451 structures, but their use or description of symbols differ from how an expert might use symbols to convey 452 the same information. Sketches that contained elements from multiple locations on the DNA Landscape 453 were coded in each location, so the sum of sketches here exceeds the 277 total sketches generated in the 454 interviews. There were no errors in the single sketch of a chromosome map, so this data point is omitted 455 here. 456 457 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 22 458 Figure 3. Example sketches that contain scale errors. Each sketch here shows an error in representing 459 the scale of a gene. Alex and Donna show a gene as corresponding to only a few base pairs, while Sheryl 460 shows a single gene taking up a large portion of a chromosome. 461 462 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 23 463 Figure 4. Example sketches that contain abstraction errors. The sketches made by Brittany and 464 Norma indicate misunderstandings of the conventional meaning of the double helix shape commonly used 465 to represent DNA. Lucy’s sketch indicates a misunderstanding of the conventions for representing genes 466 with a box and line diagram. 467 468 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 24 469 Figure 5. Example sketches that contain unconventional abstraction. Sylvia’s sketch uses 470 unconventional color coding in a representation of a replicated chromosome. Socks unconventionally 471 ascribes meaning and colors to the conventional shapes of a box and line diagram of a gene. Evelyn’s 472 sketch is unconventional in that a gene is only indicated on one chromatid in a representation of a 473 replicated chromosome. 474 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 25 475 Figure 6. Example sketches that contain both errors in scale and abstraction. Heather’s sketch 476 misrepresents the scale of a gene and misuses the conventions of the ladder shape to represent mRNA. 477 When compared to her original sketch of a chromosome, Christin’s sketch of a gene indicates a 478 misunderstanding of the size and structure of a gene. Text labels were added to Christin’s sketches for 479 clarity. Annie’s sketch indicates a misunderstanding of the size and structure of both nucleotides and 480 proteins. 481 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 26 482 Figure 7. Example sketch that is annotated in a way that reflects an understanding of scale. Piper 483 annotated the short nucleotide sequence with an ellipsis and the label “X # of [base pairs] long,” which 484 indicates she understood that genes are longer sequences than what is practical to draw in a quick sketch. 485 486 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 27 Table 1. Research participant demographics PUI R2 % Gender Man 3 3 17 Woman 8 18 74 Non-binary or gender queer 1 2 9 Year in college Sophomore 4 9 37 Junior 5 6 31 Senior 3 5 23 Master’s or Continuing Education 0 3 9 487 488 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 28 489 490 491 Table 2. Tasks completed by interview participants and alignment of interview tasks to levels of Bloom’s Taxonomy Generalized interview task Specific interview prompt Targeted Bloom’s Taxonomy level(s) Provide a verbal definition or description “What is a gene?” Remember Make a sketch “Can you sketch a gene?” Remember/Understand Explain the sketch “Why did you draw the gene that way?” Remember/Understand Explain the figure from a concept assessment item “Explain what you think about this representation of a gene.” Remember/Understand Evaluate the strengths and limitations of the figure “What about this image makes it an effective representation of a gene?” Evaluate Compare the sketch to the figure “How does your sketch of a gene compare to this representation?” Analyze Explain or draw what it would look like to “zoom into” this molecular structure “Please sketch or describe what it would look like if you “zoomed in” to a gene.” Remember/Understand Explain or draw what it would look like to “zoom out” of this molecular structure “Please sketch or describe what it would look like if you “zoomed out” of a gene.” Remember/Understand Select the answer to a concept assessment item “Please talk through how you answered this question.” Apply Translate the answer from the concept assessment item into a new sketch “Can you sketch a representation of the answer you selected?” Apply Note: Bloom’s Taxonomy levels classified based on the Visualization Blooming Tool (15) .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 29 Table 3. Codes for analyzing scale and abstraction in student sketches Code Code description Erroneous scale The participant used or described symbols in their sketch that reflected a factual error in their understanding of the size of a particular molecular structure or the scalar relationship between multiple molecular structures. Erroneous abstraction The participant used or described symbols in their sketch that reflected a misunderstanding of the visual conventions that are typically used to represent a particular molecular structure or concept. Unconventional scale The participant used or described symbols in their sketch to convey factually correct information about the size of a particular molecular structure or the scalar relationship between multiple molecular structures, but their use or description of symbols differ from how an expert might use symbols to convey the same information. Unconventional abstraction The participant used or described symbols in their sketch to correctly convey factually correct information about a particular molecular structure or concept, but their use or description of symbols differ from how an expert might use symbols to convey the same information. 492 493 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint 30 Table 4. Errors or unconventional uses of scale and abstraction in student sketches Code Number of sketches Percent of sketches Number of participants Percent of participants Scale error 47 17% 27 77% Abstraction error 80 29% 30 86% Unconventional scale 0 0% 0 0% Unconventional abstraction 27 10% 17 49% Note: These counts include a total of 14 sketches that had co-occurrences of errors in scale and errors in abstraction and 3 sketches that had co-occurrences of errors in scale and unconventional abstraction. 494 495 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 2, 2025. ; https://doi.org/10.1101/2025.02.28.640796doi: bioRxiv preprint