Edited by:
Reviewed by:
*Correspondence:
This article was submitted to the journal Frontiers in Human Neuroscience.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Functional magnetic resonance imaging was used to identify the brain-based mechanisms of uncertainty and certainty associated with answers to multiple-choice questions involving common misconceptions about electric circuits. Twenty-two scientifically novice participants (humanities and arts college students) were asked, in an fMRI study, whether or not they thought the light bulbs in images presenting electric circuits were lighted up correctly, and if they were certain or uncertain of their answers. When participants reported that they were unsure of their responses, analyses revealed significant activations in brain areas typically involved in uncertainty (anterior cingulate cortex, anterior insula cortex, and superior/dorsomedial frontal cortex) and in the left middle/superior temporal lobe. Certainty was associated with large bilateral activations in the occipital and parietal regions usually involved in visuospatial processing. Correct-and-certain answers were associated with activations that suggest a stronger mobilization of visual attention resources when compared to incorrect-and-certain answers. These findings provide insights into brain-based mechanisms of uncertainty that are activated when common misconceptions, identified as such by science education research literature, interfere in decision making in a school-like task. We also discuss the implications of these results from an educational perspective.
Answering scientific questions correctly has always been a considerable challenge for students and an important – and sometimes exclusive – indicator of academic success. One of the reasons why it is such a challenge is that students sometimes have “misconceptions” about natural phenomena that interfere with learning and, therefore, divert students from giving correct answers. Misconceptions, as typically defined in the science education research tradition, are understood as “deeply rooted” ideas or common erroneous beliefs about how nature works that are “not in harmony with the science views or are even in stark contrast to them” (
School science, as well as educational researchers, often uses multiple-choice (or true/false) questions to assess students’ performances or to diagnose misconceptions. Indeed, multiple-choice questions often include wrong answers that intentionally contain misconceptions in order to test students’ knowledge of the material. In such contexts, the production of answers becomes a matter of decision making. Indeed, decision making can be defined as the “process of making choices or reaching conclusions” (
Indeed, in previous research, we have been able to show that subjects’ expression of certainty/uncertainty about their own conceptions can have a very important effect on subsequent learning in science (
Inspired by these research efforts, particularly
This particular categorization will allow us to formulate two research questions that have the potential to help us better understand, from a neuroscientific point of view, the ordinary educational context of novice students assigned to a multiple-choice task involving misconceptions in electricity. The first question will address the issue of differences that exist between certain and uncertain answers. The results will allow us to establish interesting links between studies that have been conducted with other kinds of tasks. The second question will address the issue of the difference between cases where misconceptions most likely prevailed (OE) and cases where they were overcome (LC).
Considered as a decision-making process, the educational event of producing an answer to a multiple-choice question can be linked to other research in the field and can therefore be used as a starting point. However, according to
In most studies, the experimenters control the level of uncertainty by setting the level of ambiguity (
However, since typical school science questions about concepts are not ambiguous, and since their formulation usually excludes uncertain probabilities of events to occur, we will concentrate on studies interested in uncertainty levels that are determined by internal lacks of knowledge.
In a functional magnetic resonance imaging (fMRI) study (
Other studies have preferred to use memory recognition tasks where subjects reported different levels of confidence in their recollections of previously learned information, such as words or faces (
These studies bear, to our knowledge, the closest resemblance to school tasks that aim at assessing or developing conceptual knowledge and can better enlighten the process of students resolving a conceptual task. They will therefore serve as a basis for formulating our certainty/uncertainty hypotheses. However, since they have not tested certainty issues when misconceptions are involved, a comparison with their results will be of interest. Therefore, hypothesis No.1 (Uncertainty > Certainty) includes the activation of posterior fronto-median cortex (BA8;
Recent research efforts have argued that for many kinds of scientific learning, the executive function of inhibition might play a very important role, as in conservation problems (
In order to test our hypotheses, we have chosen to study the pedagogical context of learning about electricity because misconceptions about basic electric phenomena have been studied thoroughly in the past and are therefore well known by the science education community. We have thus developed a cognitive task involving electric circuits that was based on educational research on students’ misconceptions regarding this topic (e.g.,
Twenty-three right-handed participants took part in the study. The images from 1 participant were excluded from the analysis due to a technical problem with the response box. Therefore, the images from 22 participants (11 males, between the ages of 18 and 20;
Since we needed “novice” participants who were less likely to have prior knowledge of electricity, we recruited humanities and arts college students who had never taken optional science courses during their studies. Since individuals with anxiety disorders may have different brain activations under uncertainty in the frontal and limbic regions (
Based on a number of educational studies emphasizing the typical difficulties surrounding the understanding of simple electric circuits (e.g.,
Prior to the experiment, the bank of images of electric circuits was tested with 243 participants (with characteristics such as age, school level, and gender comparable to those of participants who were later tested in the MRI machine) in order to select a set of images that would optimize the chances of obtaining, with an equivalent set of participants, a sufficient number of certain and uncertain responses, and also to control some possibly confounding variables. Thus, the final selection of images for the cognitive task used in the fMRI optimized the possibility of obtaining balanced numbers of scientifically correct/incorrect circuits, as well as an approximately equivalent proportion of correct/incorrect answers and of certain/uncertain answers.
The final selection of stimuli used during the fMRI sessions was composed of a set of 288 electric circuits divided into four equivalent series of 72 randomly presented trials (
After they gave their written informed consent, participants were taken to a simulation room with a computer on a desk and a MRI SimulatorTM (Psychology Software Tools, Inc.). Sitting at the desktop computer, they had to read the instructions for the task and do a practice task, which was composed of 20 electric circuits similar to those used in the fMRI task. Afterward, the participants repeated the practice task, but in the MRI simulator. Participants were, at that time, explicitly informed not to move during the practice task and imaging acquisition. Immediately afterward, the task was administered in the real fMRI machine. Structural images were also obtained at the end of the four functional image series.
Imaging was performed in a Siemens 3.0 Tesla MAGNETOM Trio TIM using a 32-channel head coil. Functional images were obtained with a gradient echo EPI sequence (TR = 2000 ms, TE = 30 ms, FA = 90°, matrix size = 64 × 64, voxel size = 3 mm × 3 mm × 3 mm, number of slices = 33, slice gap = 25%, interleaved, AC-PC line orientation, whole brain scanned). The first two images were automatically eliminated by the system. Structural images were obtained with a MPRAGE sequence (TR = 2300 ms, TI = 900 ms, TE = 2.98 ms, FA = 9°, matrix size = 256 × 256, voxel size = 1 mm × 1 mm × 1 mm, number of slices = 176, interleaved, sagittal orientation). Head motion was minimized by cushions arranged around each participant’s head. Stimuli were presented with E-Prime 2.0 software (Psychology Software Tools, Inc.) via a mirror and a projection system. Subjects’ responses were collected with the Fiber Optic Button Response System (Series 1) from Psychology Software Tools, Inc.
Data analysis was performed using SPM8 (Wellcome Department of Imaging Neuroscience, London, UK). Each participant’s functional data were motion-corrected (realignment with mean image), spatially normalized (into the standard MNI space using the segmentation method in SPM8), and smoothed (using a Gaussian kernel of 8 mm FWHM). The general linear model (GLM) was used for modeling the data. More precisely, trial-related activity was modeled by convolving a vector of trial onsets with a canonical hemodynamic response function (HRF). The six movement parameters were also included in the model as regressors of no interest.
Three of the 22 subjects had one functional series with an empty condition because they were always sure (or unsure) of their answers for all the electric circuits in a series. These three series were excluded from the analysis.
Responses to a total of 5976 electric circuits were analyzed. Twenty of them (0.3%) remained unanswered by participants. Participants reported to be certain of 64.7% of their answers and uncertain of 35.0%. Behavioral task results (
Overview of the behavioral task results.
Uncertainty | Certainty | |
---|---|---|
Mean reaction time | 3911 ms | 3091 ms |
Standard deviation | 1730 ms | 1561 ms |
Total number of stimuli | 2091 | 3865 |
Number of stimuli by type of electric circuit | ||
Scientifically correct circuit | 1060 | 1750 |
Scientifically incorrect circuit | 1031 | 2115 |
Number of stimuli by students’ answer | ||
“The circuit is correct” | 945 | 1881 |
“The circuit is incorrect” | 1146 | 1984 |
Number of stimuli by value of the answer | ||
Right answer | 1030 | 2178 |
Wrong answer | 1061 | 1687 |
Number of stimuli by gender | ||
Men | 891 | 1978 |
Women | 1200 | 1887 |
Number of stimuli by series | ||
Series 1 and 2 | 1084 | 1853 |
Series 3 and 4 | 1007 | 2012 |
The stimuli for which participants claimed to be uncertain of their answers were composed of a similar number of scientifically correct and incorrect electric circuits (50.7 and 49.3%, respectively), right and wrong answers (49.3 and 50.7%), and men’s and women’s answers (42.6 and 57.4%). There was also a comparable number of stimuli that were evaluated as being correct and incorrect by participants (45.2 and 54.8%), and there were as many uncertain stimuli in the first two series of the session as there were in the last two (51.8 and 48.2%).
The stimuli for which participants claimed to be certain of their answers were also composed of a similar number of correct and incorrect electric circuits (45.3 and 54.7%, respectively), right and wrong answers (56.4 and 43.6%), and men’s and women’s answers (52.2 and 48.8%). Furthermore, there was a comparable number of stimuli that were evaluated as being correct and incorrect by participants (48.7 and 51.3%), and there were as many uncertain stimuli in series 1 and 2 as there were in series 3 and 4 (47.9 and 52.1%).
These results enable us to assume that the obtained results cannot be attributed to unbalanced quantities of particular types of answers, as would be the case if we had a prevalence of the following types of answers: masculine (vs. feminine), associated with scientifically correct circuits (vs. incorrect), answered as correct (vs. answered as incorrect), right (vs. wrong), or chronologically presented at the beginning of the task (vs. at the end of the task).
For the Uncertainty > Certainty contrast, four brain areas were significantly more activated (
Overview of the neuroimaging results (
Regions | |||||
---|---|---|---|---|---|
L middle/superior temporal gyrus (BA 21/22) | 41 | -63 | -39 | 3 | 5.36 |
R superior frontal gyrus (BA 8/9) | 34 | 9 | 54 | 45 | 5.32 |
LR anterior cingular cortex (BA 24/32) | 59 | -3 | 36 | 21 | 4.86 |
L inferior frontal gyrus/superior temporal gyrus and insula (BA 47/38/13) – into the lateral sulcus | 27 | -36 | 15 | -12 | 4.49 |
LR from middle/inferior occipital gyrus and inferior temporal gyrus (BA 19/18/37, L peak) to angular gyrus (BA 39) and superior parietal lobule (BA 7/19) | 1545 | -45 | -72 | -3 | 8.40 |
LR inferior parietal lobule and postcentral gyrus (BA 40/2) | 83 | 63 | -24 | 42 | 6.31 |
R inferior/middle frontal gyrus (BA 44/8), insula (BA 13), and precentral gyrus (BA 4/6) | 40 | 42 | -3 | 24 | 5.27 |
L insula (BA 13) and precentral gyrus (BA 4/6) | 38 | -33 | 0 | 18 | 5.19 |
L insula (BA 13) and precentral gyrus (BA 4/6) | 22 | -48 | 0 | 6 | 4.62 |
For the Certainty > Uncertainty contrast, a large bilateral activation, beginning in the middle/inferior occipital gyrus and the inferior temporal gyrus and ending in the angular gyrus and superior parietal lobule, was observed when the participants were certain of their answers compared to when they were uncertain (
For the LC > OE contrast, five brain areas were significantly more activated (
Overview of the neuroimaging results (
Regions | |||||
---|---|---|---|---|---|
L superior/inferior parietal lobule (L intraparietal sulcus; BA 19/7) | 26 | -27 | -48 | 48 | 5.64 |
R premotor cortex (BA 6) | 87 | 21 | -9 | 45 | 5.51 |
R superior/inferior parietal lobule (R intraparietal sulcus; BA 19/7) | 102 | 33 | -51 | 63 | 5.50 |
R inferior temporal gyrus (fusiform gyrus; BA 37) | 34 | 48 | -48 | -9 | 4.91 |
R precentral/postcentral gyrus (motor cortex; BA 4/3) | 43 | 51 | -21 | 51 | 4.65 |
L precentral/postcentral gyrus (motor cortex; BA 4/3) | 72 | -42 | -18 | 57 | 5.78 |
All contrasts related to Certainty > Uncertainty (LC > UE; LC > LD; OE > UE; OE > LD) showed patterns of activation that we also found in the general Certainty > Uncertainty contrast (i.e., occipito-parietal activations). None of the other possible contrasts between the four response categories (UE > LD; LD > UE; LD > OE; UE > LC) recorded significant activation, except for UE > LC. This particular contrast recorded a significant activation of the anterior cingulate cortex. (MNI coordinates:
Our hypothesis for Uncertainty > Certainty included the activation of the anterior cingulate, the superior frontal gyrus and the posterior fronto-median cortex (BA8; with particular attention to medial temporal activations). It appears that much of this hypothesis is confirmed, since these regions were activated, with few unexpected activations (see below). We therefore believe that our results are in line with results obtained in other studies that were looking for uncertainty caused by lacks of knowledge.
Indeed, the ACC which has been reported in a number of related studies (
Also, in compliance with our hypothesis, the superior frontal gyrus and dorsomedial prefrontal cortex were activated. The superior frontal gyrus has often been reported in other studies about decision making under uncertainty (
Another region that is more activated under uncertainty is the left superior temporal gyrus, extending to the middle temporal gyrus. Contrary to the brain areas previously discussed (ACC, insula, and superior/dorsomedial frontal cortex), the activation of this region has been rarely reported in studies about decision making under uncertainty, and when it has been, the results were contradictory. For example,
We also recorded activation of the insula (extending to the inferior/middle frontal gyrus and the precentral gyrus). This area has also been activated in a number of studies about decision making under uncertainty (
We did not formulate explicit hypotheses about the Certainty > Uncertainty contrast, since the studies we cited in the previous section did not extensively discuss the brain-based mechanisms related to certainty. There are still, however, a few studies that focus on the neural correlates of certainty associated with decision. For example, some researchers have studied the brain-based mechanisms of confidence in recognition memory (e.g.,
It is true that our task presented images that were similar to realistic photographs of electric circuits and did not refer to abstract representations of circuits, such as mere schemas with straight lines, nor did it not use sentences to describe scientific phenomena or conceptions. Indeed, we believe that if our stimuli had been induced with words instead of images, by stating for example that “one wire is sufficient to light a bulb,” it would have been highly likely that visuospatial regions might not have been recruited as much. It does not appear unreasonable, however, to suggest that scientific competency (at least with electric circuits) might be associated with visuospatial treatment. Therefore, dealing with real-life tasks (like school lab experiments) and complex problems that necessitate visual treatment could improve the ease with which students learn about electrical concepts.
In accordance with previous research on the differences between novices and experts in scientific tasks involving misconceptions (see above Section “Difference Between Over-Estimation and Legitimate Certainty”), our second hypothesis suggested that contrasts between LC and OE (LC > OE) would show activations of the ACC and the dorsolateral and ventrolateral cortices. This hypothesis could not be confirmed. Typical inhibition mechanisms did not show any significant activation for
The activation of the left and right intraparietal sulcus that was recorded is usually associated with number processing (
From an educational perspective, we first believe that presented results are rather encouraging because they suggest that resolutions of multiple-choice educational tasks could be understood as decision-making processes. Second, since during our task we never gave any negative feedback to the participants, we believe that activations that occur during uncertainty can be associated with an internal conceptual conflict between competing conceptions. It is also possible that the activation of the left temporal lobe, and the corresponding suggestion that inner language was used, could be an indication of a conflict that would require a heavier dialectic process to be resolved. We believe that these are interesting results because similar interpretations have been proposed, but for differences between experts and novices (
For the “certainty” category of responses, we found large posterior activations, usually associated with visual and spatial processing. Our interpretation suggests that scientific conceptions about electricity might be grounded in the visuospatial circuits of the brain. Even if other means should not be discarded as interesting ways to develop scientific conceptions, we hypothesize, that visuospatial training with real (or realistic) tasks (such as laboratory tasks or working with realistic images), might lead to more confidence when tackling tasks like the one we used in our study. We believe that this suggestion is quite in line with
Our second hypothesis was that we would find typical inhibition activations in the contrast between correct-and-certain answers and incorrect-and-certain ones (LC > OE). In this setup, we hypothesized that correct-and-certain answers would be typical of expert answers. However, we were not able to show typical expert or inhibitive activations. Instead, most recorded activations could be interpreted as the use of left and right fingers, and possibly the mobilization of identification and number processing. Based on these results, we can hypothesize that correct answers might require a more thorough examination of our electrical circuits. Indeed, some were rather complex and might have necessitated a greater number of verifications (identification and enumeration of all different parts) and therefore the mobilization of visual attention resources. It can also be suggested that in order to answer correctly to scientific questions, novices recruited resources that differed from the ones experts would recruit.
This research can be considered as an effort to link neuroscientific and educational knowledge through the use of an authentic educational context involving the resolution of a multiple-choice task (considered as a decision-making process). We believe that this effort has been fruitful because the recorded activations about uncertainty/certainty were typical of decision-making processes that involved uncertainty caused by “lacks of knowledge.” Therefore we believe that many of the extended knowledge elements about decision making might prove useful in the long run to better understand school performance and failure. But we also believe that much more work has to be done in order to better understand the differences between the production of correct answers and expertise, and also the origins and the function of uncertainty in learning. In science, uncertainty can indeed be the driving force of knowledge development, but in some cases, it can also be paralyzing. Thus expert uncertainty might be at least somewhat different from novice uncertainty.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
This research was made possible with the help of the Social Sciences and Humanities Research Council of Canada (SSHRC, grant number 365011) and by the Canada Foundation for Innovation (CFI, grant number 12751). Special thanks to Frédérick Fortin for his help with the data analysis.