Elsevier

Cognition

Volume 192, November 2019, 104001
Cognition

Original Articles
Perception and conception in understanding evolutionary trees

https://doi.org/10.1016/j.cognition.2019.06.013Get rights and content

Abstract

Relationships depicted in evolutionary trees depend solely on levels of most recent common ancestry. Integrating discipline-based education research in biology with perceptual/cognitive psychology, the authors predicted, however, that the Gestalt principles of perceptual grouping would affect how students interpret these relationships. Experiment 1 (N = 93) found that students segment 6–9 branch trees in accordance with the Gestalt principle of connectedness. Experiment 2 (N = 310) found that students in introductory through advanced biology classes predominantly believed, incorrectly, that the evolutionary relationships among a set of target taxa differed in two trees because the grouping of those taxa differed. Experiment 3 (N = 99) found that students from these same classes were more likely to make inferences consistent with the depicted evolutionary relationships when Gestalt grouping supported those inferences. The authors discuss implications for improving students’ understanding of cladograms.

Introduction

Discipline-based education research (DBER) investigates how post-secondary students (a) learn concepts, practices, and ways of thinking and (b) develop problem-solving expertise in science and engineering (National Research Council [NRC], 2012). Because DBER is focused on thinking and learning that is deeply embedded in a content domain, it is necessarily guided by expert knowledge of that domain. It is important to keep in mind, however, that all instances of problem solving, reasoning, and learning are accomplished with the same information processing system, which has particular, specifically human, characteristics and ways of functioning (e.g., Simon, 1978). Understanding this system is the goal of cognitive and perceptual psychology.

Clearly, then, to obtain a full understanding of thinking and learning in science domains, researchers must consider the contributions of both domain-specific knowledge and general aspects of cognitive processing (also see Coley & Tanner, 2015). The present studies leverage a collaboration between DBER and cognitive/perceptual psychology (also see NRC, 2012) to (a) reveal insights concerning abstract reasoning in evolutionary biology and (b) inform cognitive and perceptual psychologists about the applicability of laboratory-defined principles to reasoning with authentic scientific diagrams.

A burgeoning area of DBER in evolutionary biology concerns students’ ability to interpret and reason with the information depicted in evolutionary trees—i.e., to engage in tree thinking. Consider the phylogenetic tree—specifically, a cladogram1—shown in Fig. 1. Cladograms depict hypothesized evolutionary relationships in terms of nested clades. A clade is a group of taxa consisting of a most recent common ancestor and all its descendants. The cladogram in Fig. 1 shows, for example, that, among the nine taxa depicted, skunk and raccoon comprise a clade, as do skunk, raccoon, and dog. The nesting of these two clades indicates that raccoon is more closely related to skunk than to dog because it shares a more recent common ancestor (MRCA) with the former than the latter taxon. Using similar reasoning, moles are equally closely related to rabbits and raccoons because no two of these three taxa share a MRCA with each other than with the third taxon. The ability to engage in tree thinking is a critical component of science literacy and thus an important part of STEM (science, technology, engineering, and math) education (Baum and Smith, 2013, Novick and Catley, 2013, Thanukos, 2009). (Interested readers may consult Novick, Catley, & Schreiber, 2012, for a more in-depth, research-based primer on tree thinking.)

Tree thinking is an ideal testbed for integrating DBER and cognitive/perceptual psychology. Evolutionary relatedness, which depends on levels of most recent common ancestry (i.e., relative recency of common ancestry) and is the key concept depicted in cladograms (e.g., Brower, 2016), is a threshold concept (Meyer & Land, 2003) in evolutionary biology. It is the conceptual foundation for mastering tree thinking (e.g., Baum and Smith, 2013, Hennig, 1966, Novick and Catley, 2013), and research shows that it is difficult for students to learn (e.g., Dees et al., 2014, Novick and Catley, 2013, Novick and Catley, 2016, Novick et al., 2014, Phillips et al., 2012). The extent of this difficulty suggests that it may stem from both the nature of the conceptual knowledge to be acquired and constraints due to the nature of the human information processing system.

Consider, again, the cladogram in Fig. 1, which represents a miniscule portion of the unimaginably large Tree of Life (ToL): It depicts relationships among only 5 of the more than 5000 extant mammalian species (Wilson & Reeder, 2005), 2 of the more than 16,000 extant non-avian reptile species (Lee, Reeder, Slowinski, & Lawson, 2004), 1 of the more than 28,000 fish species (Nelson, 2006), and 1 of the more than 7000 echinoderm species (Brusca & Brusca, 2003). The complete ToL also includes, for example, more than 6 million extant arthropod species (Stork, McBroom, Gely, & Hamilton, 2015), none of which are included in this cladogram, as well as billions of extinct species (McKinney, 1997). By focusing on a tiny portion of the ToL, a particular cladogram necessarily calls attention to the pattern of relationships among certain taxa while backgrounding the relationships of these taxa to others that are not displayed.

When interpreting cladograms, biologists focus on the evolutionary relationships that are depicted—i.e., patterns of most recent common ancestry. How these relationships are represented in cladograms and interpreted in an evolutionary context constitute domain-specific content knowledge that must be learned through instruction. Importantly, these relationships do not change depending on which taxa are included in a cladogram because there is only one ToL, whose structure was set by how evolution has proceeded over the past approximately 3.5 billion years. We contend, however, that these are not the only relationships that are illustrated in cladograms. There are also perceptually-defined grouping relationships, which are specified by the Gestalt principles of grouping (e.g., Wertheimer, 1923/1958). These relationships are readily apparent to all observers and do not depend on instruction. They do, however, depend on which subset of taxa from the ToL are included in the cladogram.

Novick and Catley (2016) recently hypothesized that the fundamental difficulty students need to overcome to acquire expertise in tree thinking may be understanding that any specific cladogram is a subset of the complete, unimaginably large ToL. They suggested that students reify the particular groupings of taxa they see and fail to appreciate that those groupings are largely an artifact of the specific subset of taxa that happens to be included in the cladogram. We view these groupings as artifactual because it is easy to change the apparent groupings by changing the taxa that are included. As noted, this does not change the evolutionary relationships depicted.

For example, Novick and Catley (2016) showed students in an introductory biology class for science majors the cladogram in Fig. 1 after they had received instruction in tree thinking. Students were asked whether (a) moles are more closely related to rabbits than to raccoons, (b) moles are more closely related to raccoons than to rabbits, or (c) rabbits, moles, and raccoons are all equally closely related to each other. Students overwhelmingly (90%) selected option (a). The correct answer is (c).

To see why this is the case, consider Fig. 2, which shows two different revisions to the cladogram in Fig. 1, both of which preserve the evolutionary relationships among rabbits, moles, and raccoons depicted in that cladogram. In Fig. 2a, skunk and dog have been removed, which makes it easy to see that the three taxa in question form a polytomy—a set of three or more taxa with the same most recent common ancestor. Thus, rabbits, moles, and raccoons are equally closely related to each other. When skunk and dog are included in the cladogram, as in Fig. 1, however, what stands out perceptually is that raccoons share a MRCA with skunks (and dogs) than with rabbits and moles. However, as shown in Fig. 2b, one could similarly group rabbits and moles with other taxa too, just by adding one or more taxa with which they each share a MRCA: e.g., pika for rabbit and hedgehog for mole. Critically, adding or deleting taxa with which rabbits, moles, and raccoons are each more closely related does not change the relationship among those three taxa, which depends solely on the location of their most recent common ancestor.

A fundamental problem identified by the Gestalt psychologists in the first quarter of the 20th century is that complex visual scenes must be segmented by the viewer into (a) discrete objects that constitute the figure, which will become the focus of attention, and (b) the remainder of the scene, referred to as the ground (e.g., Koffka, 1935, Köhler, 1926, Wertheimer, 1923; see Wagemans, Elder, et al., 2012, for a review). These psychologists further described several perceptual principles (e.g., proximity, good continuation; see Fig. 3a-c) that occur without volitional control and that govern this segmentation. Subsequent researchers have described additional grouping principles, of which element connectedness is most relevant here (see Fig. 3d; Humphreys and Riddoch, 1993, Palmer and Beck, 2007, Palmer and Rock, 1994). More recent theoretical work has attempted to formulate computational models of perceptual organization (see Wagemans, Feldman, et al., 2012, for a review). Notable among these efforts is that of Froyen, Feldman, and Singh (2015), whose Bayesian Hierarchical Grouping model provides a unified account of multiple perceptual organization phenomena without assuming the Gestalt principles as premises. In the discussion that follows, we focus on the grouping principles per se because that level of description is most relevant for our research.

Proximity and element connectedness, which seem critical for determining the perceptual organization of hierarchical trees, are among the strongest grouping principles (e.g., Quinlan and Wilton, 1998, Zhao et al., 2016). The principle of proximity (see Fig. 3b) is that, all other things being equal, two or more elements that are near each other will be grouped into a single unit. The principle of element connectedness (see Fig. 3d) is the tendency for two or more distinct elements to be grouped into a single unit when they are connected together.

Perceptual grouping is a critical part of visual processing, and thus a basic aspect of the human information processing system, because it yields the particular objects that viewers experience. These objects then become the units of cognition. Accordingly, an important area of research has been to identify the cognitive consequences of having settled on a particular perceptual organization of a scene into parts. Researchers have found, for example, that the Gestalt principles affect what gets stored in visual working memory (Woodman, Vecera, & Luck, 2003) and how spatial relations are represented and remembered (Coren and Girgus, 1980, Maki, 1981, McNamara, 1986, Stevens and Coupe, 1978).

Consistent with Pinker’s (1990) theory of graph comprehension, there is also evidence that the Gestalt principles of similarity, proximity, and connectedness affect how students describe the information depicted in bar and line graphs (Ali and Peebles, 2013, Shah et al., 1999, Zacks and Tverskv, 1999). It is notable that the task in these studies was to verbally describe the presented graphs. Students were not asked to reason from the information presented or to extract specific relationships from the graphs. In contrast, Landy and Goldstone (2007) examined the effect of Gestalt grouping on students’ ability to make a complex cognitive judgment. They found that students’ judgments of the validity of algebraic equations were affected by how the elements were grouped perceptually, even though equation validity depends solely on the mathematical principle of order of operations (e.g.,

Yes or No?).

In research that is even more closely related to the present topic, Novick and Catley, 2007, Novick and Catley, 2013, Novick et al., 2010 and Catley, Novick, and Funk (2012) compared students’ understanding of and ability to engage in tree thinking with cladograms drawn in two alternative, structurally equivalent formats (see Fig. 4). This research demonstrated that (a) students have greater difficulty interpreting trees in the diagonal (Fig. 4a) than the rectangular (Fig. 4b) format, and (b) this difficulty is due to the Gestalt principle of good continuation (Shimaya, 1997, van Tuijl, 1980). This principle states that a continuous line is (or represents) a single entity; thus, Fig. 3c is segmented into a circle and a thick vertical line. Because continuous lines in diagonal format cladograms often represent multiple hierarchical relationships based on ancestry, this perceptual principle interferes with students’ ability to extract the critical structural and conceptual information regarding levels of most recent common ancestry from such cladograms.

The rectangular cladogram format is most commonly used in the professional literature in evolutionary biology (Novick & Catley, 2007) and, given the results of the studies discussed above, it is now also the dominant format used in textbooks for the introductory biology class for science majors. Although the Gestalt principle of good continuation does not interfere with students’ interpretations of such cladograms, we hypothesize that the principles of proximity and, especially, element connectedness play an important role in leading students to misinterpret trees in the rectangular format.

Our research testing this hypothesis goes beyond previous research on the link between perception and conception using complex stimuli in several ways. First, rather than examining open-ended description, we investigated the effects of Gestalt grouping on students’ ability to extract specific information from authentic scientific diagrams and to make inferences using the information depicted. Although it is reasonable to expect that how students describe multi-part diagrams will affect their ability to make more complex judgments, such as those just mentioned, this need not be the case. For example, when asked to attend to a particular part of a complex diagram to extract specific relational information or to make an inference, students may be able to do so even if they did not focus on that part of the diagram when providing a description.

Landy and Goldstone (2007) investigated reasoning, but our studies differ from theirs in two important ways. First, our stimuli are diagrams rather than mathematical equations. More importantly, our manipulation of grouping compares authentic versions of the scientific diagrams. As we discuss in Section 3, the different groupings naturally result as a function of which taxa are selected from the complete ToL. In contrast, Landy and Goldstone’s manipulation involved presenting mathematical equations in ways that were specifically designed to trick students into adding before multiplying, contrary to the mathematical rule of order of operations and to how most formulae are presented to students.

Finally, we manipulated grouping within a single type of diagram—rectangular format cladograms—rather than comparing performance across different types of diagrams, with different grouping conventions, as was generally the case in the research using bar and line graphs. Shah et al.’s (1999) Experiment 3 was an exception, but their within-graph-type manipulation relied on comparing typical versus atypical depictions for each graph type, and their results indicated that students had difficulty understanding the atypical depictions. Our within-diagram-type manipulation of grouping does not have this drawback because it is a natural result of the subset of taxa that are chosen for inclusion in the cladogram.

This article reports the results of three experiments at the intersection of cognitive/perceptual psychology and DBER. All experiments were approved by the relevant IRB committees, and students provided informed consent before participating.

The three experiments test two interrelated hypotheses. The first hypothesis is that students segment hierarchical diagrams, of which cladograms from evolutionary biology are an important example, in ways that are predictable given the Gestalt principles of grouping. Because this hypothesis concerns perceptual structure, independent of the content to which it is applied, our test of it in Experiment 1 used rectangular trees with unlabeled branches. Students were shown two alternative ways each tree could be divided into two groups and had to indicate which subdivision seemed more natural. One subdivision was consistent with Gestalt principles of grouping; the other was not. We predicted that students would select the Gestalt-consistent subdivision.

The second hypothesis is that how students interpret the pattern of evolutionary relationships in cladograms is affected by the perceptual grouping of the taxa. In particular, we hypothesize that part of the reason why students have difficulty with tree-thinking tasks is that when evolutionary structure and Gestalt grouping conflict, students are likely to respond based on Gestalt grouping. We tested this hypothesis in two experiments that employed very different tree-thinking tasks. In Experiment 2, students received pairs of cladograms for which the pattern of evolutionary relationships among three target taxa was identical but the perceptual grouping relationships differed. We predicted that students’ interpretations of the evolutionary relationships would be consistent with the Gestalt grouping rather than the levels of common ancestry.

In Experiment 3, students received individual cladograms and were asked to make inferences about a biological property of one taxon based on such information provided about two other taxa. Inference based on evolutionary relationships is a key tree-thinking skill for students to learn (Novick and Catley, 2013, Novick and Catley, 2016) because it is critical in both basic and applied biology. Our version of the task drew on the large cognitive science literature on biological inferences, which we discuss in the introduction to Experiment 3. In our task, inferences should be based on recency of common ancestry (i.e., evolutionary relationship), but they could also be based on Gestalt grouping. We manipulated whether the Gestalt grouping relations facilitated or hindered making the appropriate (relationship-based) inference.

Section snippets

Experiment 1

Students received pairs of diagrams (e.g., see Fig. 5) in which each pair showed two alternative ways of dividing a hierarchical tree into two parts. Their task was to indicate which subdivision of each tree seems “most natural or most likely.” This task has been used effectively to investigate how Gestalt grouping principles affect segmentation of line drawings into parts (Novick and Catley, 2007, van Tuijl, 1980). For the two pairs of trees in Fig. 5, the principle of element connectedness

Experiment 2

Experiment 2 investigated whether the strong grouping effects identified in Experiment 1 influence how students interpret evolutionary relationships in cladograms. In particular, we asked whether students could recognize that the evolutionary relationships among a set of three target taxa were the same in two cladograms even though the perceptual grouping of those taxa differed: In one cladogram, the three target taxa appeared to belong to the same perceptual group; in the other cladogram, the

Overview

It is well known that students often evaluate evolutionary relatedness based on the distance between the taxa of interest (e.g., Gregory, 2008, Meir et al., 2007, Novick and Catley, 2013). Consider, for example, the cladogram shown in Fig. 6B. Students are likely to think this cladogram shows that platypuses are more closely related to ducks than to wolves because the platypus is physically closer to the duck than to the wolf. Horizontally, the platypus is next to the duck but three taxa away

General discussion

The ability to interpret and reason from the evolutionary relationships shown in cladograms—hierarchical trees that depict those relationships in terms of levels of most recent common ancestry—is an important aspect of science literacy (Baum and Smith, 2013, Novick and Catley, 2013, Thanukos, 2009). Unfortunately, college biology students have considerable difficulty engaging in such tree thinking even after (sometimes extensive) instruction in that topic (e.g., Catley et al., 2012, Dees et

Acknowledgments

We would like to thank Shannon Hurlston for her help with (a) drawing the stimuli for Experiment 1, (b) writing the REDCap programs to collect the data for all three studies, (c) editing the data files exported from REDCap to put them into a usable form for analysis, and (d) coding students’ written explanations. Kyra Frank also helped with coding the explanations. We thank Duane Watson for helpful discussion of the comparison questions to use in Experiment 2 and Sun-Joo Cho for helping with

References (67)

  • A.V.Z. Brower

    What is a cladogram and what is not?

    Cladistics

    (2016)
  • R. Brusca et al.

    Invertebrates

    (2003)
  • K.M. Catley et al.

    The promise and challenges of introducing tree thinking into evolution education

  • J.D. Coley et al.

    Relations between intuitive biological thinking and biological misconceptions in biology majors and nonmajors

    CBE–Life Sciences Education

    (2015)
  • S. Coren et al.

    Principles of perceptual organization and spatial distortion: The Gestalt illusions

    Journal of Experimental Psychology: Human Perception and Performance

    (1980)
  • C. Darwin

    On the origin of species by means of natural selection

    (1859)
  • J. Dees et al.

    Student interpretations of phylogenetic trees in an introductory biology course

    CBE—Life Sciences Education

    (2014)
  • V. Froyen et al.

    Bayesian hierarchical grouping: Perceptual grouping as mixture estimation

    Psychological Review

    (2015)
  • T.R. Gregory

    Understanding evolutionary trees

    Evolution: Education and Outreach

    (2008)
  • W. Hennig

    Phylogenetic systematics

    (1966)
  • G.W. Humphreys et al.

    Interactions between object and space systems revealed through neuropsychology

  • C.R. Hynd et al.

    The role of instructional variables in conceptual change in high school physics topics

    Journal of Research in Science Teaching

    (1994)
  • B.D. Jee et al.

    Drawing on experience: How domain knowledge is reflected in sketches of scientific structures and processes

    Research in Science Education

    (2014)
  • D. Kelemen

    Why are rocks pointy? Children’s preference for teleological explanations of the natural world

    Developmental psychology

    (1999)
  • D. Kelemen et al.

    Professional physical scientists display tenacious teleological tendencies: Purpose-based reasoning as a cognitive default

    Journal of Experimental Psychology: General

    (2013)
  • K. Koffka

    Principles of gestalt psychology

    (1935)
  • W. Köhler

    An aspect of Gestalt psychology

  • P. Kowalski et al.

    The effect of refuting misconceptions in the introductory psychology class

    Teaching of Psychology

    (2009)
  • D. Landy et al.

    How abstract is symbolic thought?

    Journal of Experimental Psychology: Learning, Memory, and Cognition

    (2007)
  • M.S.Y. Lee et al.

    Resolving reptile relationships: Molecular and morphological Markers

  • R.H. Maki

    Categorization and distance effects with spatial linear orders

    Journal of Experimental Psychology: Human Learning and Memory

    (1981)
  • M.L. McKinney

    How do rare species avoid extinction? A paleontological view

  • E. Meir et al.

    College students’ misconceptions about evolutionary trees

    The American Biology Teacher Online

    (2007)
  • View full text