How Children Seek Out Information from Technological and Human Informants Judith H. Danovitch (j.danovitch@louisville.edu) Department of Psychological & Brain Sciences, 317 Life Sciences, University of Louisville Louisville, KY 40292 USA Nicholaus S. Noles (n.noles@louisville.edu) Department of Psychological & Brain Sciences, 317 Life Sciences, University of Louisville Louisville, KY 40292 USA Patrick Shafto (p.shafto@louisville.edu) Department of Psychological & Brain Sciences, 317 Life Sciences, University of Louisville Louisville, KY 40292 USA Abstract enormous amount of information from nearly anywhere Members of the current generation of young children have in the world. Yet, despite the fact that nearly 43% of the been exposed to technological informants, primarily world’s population now has access to the Internet consisting of devices that search the Internet for (http://www.internetworldstats.com/stats.htm), little is information, nearly since birth. However, little is known known about how access to the Internet influences the about how young children explore information using these development of information-seeking behaviors. digital sources. To address this issue, 30 preschool children One issue that has been of interest to both popular generated questions about unfamiliar animals that were to commentators and researchers is whether Internet use be answered by either a human or technological informant (i.e., an Internet search program). Children also completed has consequences for curiosity and exploration. Some a measure of biological and psychological attributions to commentators have argued that access to information different types of information sources. Overall, children via technology and the Internet is diminishing our generated similar numbers of questions for each informant, cognitive capacity and intelligence (e.g., Bauerlein, and a similar proportion of their questions were causal in 2008; Carr, 2010). These concerns extend to include nature. Children also attributed few biological and effects on motivation to seek out information and psychological characteristics to the Internet search acquire meaningful understanding (i.e., by providing a program. This suggests that, despite understanding that technological devices share few biological and quick answer, technological informants might stifle psychological properties with people, young children seek exploration or skepticism). There is also emerging out information in similar ways from human and evidence that, at least among adults, obtaining technological information sources. information via the Internet may lead to overconfidence Keywords: technology; information; questions in one’s own knowledge (Fisher, Goddu & Keil, 2015; Ward, 2013), potentially diminishing motivation to learn more. That said, there have been no studies As human beings, we have a natural propensity to ask examining how interactions with technological questions and seek out information about the world informants influence curiosity and exploration in around us. If the answers to our questions are not children during a period when their epistemological readily observable, we look for answers from other concepts are still rapidly changing and they are sources. For the vast majority of human history, these becoming more aware of the limitations of their own sources have consisted primarily of other people (with knowledge (Mills & Keil, 2004). the later addition of books, radio, and other media). The study described here examines how consulting a Seeking out information sources often requires time and person or the Internet as an information source affects effort, and even then the answer they provide might be children’s information-seeking behaviors. Children in “I don’t know.” many modern communities have been exposed to However, in the past few decades, there has been a technological devices that can access the Internet nearly dramatic change in how people look for answers to from birth, and they frequently observe adults using their questions. With the advent of the Internet, these devices to obtain information. Anecdotally, questions that would have required consulting an parents report that their young children ask them to expert, or a trip to the library, can now be answered look up information using Google or other search within milliseconds by consulting an Internet search engines – although they do so even for questions that engine. In addition, devices such as smartphones have cannot be answered by information available via the made it possible to quickly and easily access an Internet (e.g., “is there a pizza in the freezer?” Richler, 407 2015). Thus, comparing the questions children direct to instance, although young children understand the human and technological sources can provide an insight biological and psychological differences between into how children use technological sources to obtain humans and computers (Scaife & van Duuren, 1995; information and their beliefs about the kinds of Mikropoulos, Misailidi, & Bonoti, 2003), they may information that can be obtained from human or have difficulty understanding the extent of a computer’s technological sources. information storage capacity (Subrahmanyan, Gelman, Although, to the best of our knowledge, there is no & Lafosse, 2002). That said, because much of the research looking at the questions children ask of existing research took place more than a decade ago, technological informants, there is ample existing these studies did not make any mention of the Internet. research examining the developmental trajectory of The Internet provides access to a vast amount of children’s information-seeking via questions. By age 4, information and search engines are much more children are more likely to ask questions of more interactive and selective than previous generations of knowledgeable informants than less knowledgeable technology, and even children with experience using ones (e.g., Birch, Vauthier, & Bloom, 2008; Koenig, the Internet seem to have difficulty understanding its Clement, & Harris, 2004), and they are capable of structure and complexity until late elementary school directing their questions to individuals with appropriate (Yan, 2005, 2006, 2009). Thus, our current study not background knowledge and expertise (Lutz & Keil, only provides an important update to previous work, 2002). Thus, by preschool, children are adept at but it also investigates whether children attribute differentiating between different types of information pedagogical capacities to Internet search engines. sources. In addition, by age 5, children are capable of Method formulating questions that will allow them to solve Participants problems (Mills et al., 2010). As Greif and colleagues Thirty children ranging from 4.36 to 5.89 years (Mage (2006) found, children tailor their questions to the topic = 4.89, 16 males) participated at preschools and at hand and seek out different kinds of information kindergartens in an urban area. The majority of the about unfamiliar animals versus artifacts. If, for the children were identified by their parents as Caucasian- purposes of obtaining information, children treat the American and non-Hispanic. Children were interviewed Internet search engine as having the same capabilities individually by an experimenter in a quiet area of their (e.g., ability to select the most relevant information) as school. a human, then we would expect children to show similar levels of curiosity (e.g., rates of question asking) when seeking out information from a person or Materials & Procedure an Internet search engine. However, if children view the The procedure was loosely based on the paradigm Internet as having a greater or lesser capacity to obtain developed by Greif and colleagues (2006), where information or formulate answers, then their rate of children encounter unfamiliar animals and are information seeking should differ. encouraged to generate questions about each animal. Furthermore, our study examines whether children Children interacted with two informants: a person and direct different types of questions to a human or an Internet search engine. Because questions are technological informant. Young children are highly typically presented to human and technological motivated to seek out causal explanations, and they ask informants in different ways (e.g., verbal vs. typed on a appropriate “how” and “why” questions in order to keyboard) and this might affect children’s behavior, the obtain these explanations (Frazier, Wellman, & experimenter “interacted” with both informants via a Gelman, 2009). Nevertheless, they may view the laptop computer with a 15-inch screen. Information was Internet as a good source of detailed factual submitted to both informants by typing. information, but not necessarily as a good source of The Internet search engine was represented using a causal explanations, which require additional synthesis window labeled “search” that contained a magnifying and understanding to generate. To address this glass icon (see Figure 1a). The human informant was possibility, our study examines the proportion of presented in a schematically similar manner in a questions children ask of each informant that seek out window labeled “chat” (see Figure 1b). The graphic in causal information. this window was a silhouette of a person. Both Finally, we are interested in how children’s windows featured an editable text area where the information-seeking behaviors are related to the ways experimenter could type in a question with a question in which they conceptualize human and technological mark button to the right, which the experimenter hit to sources. Some insight into this issue can be gained transmit the question to the informant. from existing research examining how children conceptualize computers and their capabilities. For 408 Other materials included full color images of each of technological informant, when the button was pushed, the four target animals (pangolin, colugo, echidna, and the magnifying glass graphic disappeared and was tarsier) printed on separate sheets of paper. replaced by a large rotating hourglass. After rotating for 5 seconds, the hourglass disappeared and was replaced by an image representing the program’s response to the query (e.g., an image of a rabbit in response to the query “what animal eats carrots?”). The experimenter also pointed at each image and said its name (e.g., “rabbit”) after it appeared. The first informant’s response remained on screen while the experimenter queried the other informant and that informant’s answer appeared on the screen. Thus, both responses were available at the end of each trial. Children were then asked to state the correct answer to Figure 1a: Screenshot of desktop with search window each question. The order in which the informants were open. queried was counterbalanced so that half of the participants always saw the human informant answer the question first, and the other half always saw the technological informant answer first. Following the familiarization trials, the experimenter cleared both windows from the screen and participants were presented with the question trials. Question Trials The experimenter introduced the first pair of question trials by telling the child that they would have a chance to learn more about some new animals by asking the person questions. She then opened the “chat” window only, which appeared in the Figure 1b: Screenshot of desktop with chat window center of the screen. Children were instructed to tell the open. The video of the person used in the experimenter their questions about the animal and the familiarization trials was replaced by a silhouette image experimenter would type the questions. Children were in the question trials. told that they could ask as many questions as they wanted, and that they would receive the answers later. Procedure The experimenter began each trial by placing a photo At the beginning of the experimental session, the of an animal on the table in front of the child and experimenter informed children that they would be asking: “Do you know what this is called?” If the child learning about animals from two different sources and answered that they did not know, the experimenter explained that one source was “a computer program continued by stating “It’s a [animal name]. What that can look for answers to questions on the Internet” questions do you want to ask the person about the and the other source was “a live video chat with a [animal]?” If the child named the animal incorrectly, person who lives in another city.” During the the experimenter corrected them by introducing the introduction, the experimenter opened and displayed animal’s name. Each time the child asked a question, the corresponding windows for each source. the experimenter repeated the question while typing it into the window on the computer screen. The Familiarization Trials Each session continued with 2 experimenter then submitted the question to the familiarization trials that involved questions about information source, and the child had the opportunity to information familiar to young children (e.g., what ask another question. There was no limit on the number animal says “moo”?). The experimenter read each of questions children could generate. However, if the question aloud as she typed the question into each child paused for more than 10 seconds, the informant’s text box and each informant gave an experimenter asked the child if s/he had any more answer in turn. For the human informant, when the questions about the animal. If the child did not generate question mark button was pressed, it was replaced with additional questions, the trial ended. video of an adult male who presented evidence by Following the second trial, the experimenter cleared looking down (off screen) for approximately 5 seconds. the screen and stated that now the child would be He then looked back up as he presented his response by learning about animals from the computer program. She holding up an image printed on a sheet of paper. For the then followed the same procedure as with the first set of 409 trials, entering the child’s question in the “search” proportion of causal questions children asked of each window instead. informant, t(24) = .998, p = .328 (see Figure 2). The order in which the child encountered the informants (person or computer program first) and the order in which the animals were presented was 120   counterbalanced between subjects. #  of  questions  asked   100   Attribution Trials The goal of this task was to 80   examine children’s intuitions about the biological and psychological nature of human and technological 60   entities. Children were instructed to answer yes or no to a series of questions about four target items presented 40   as photos on cards: a person (represented by an image 20   of the man in the chat window from the familiarization trials), a computer program (represented by an image of 0   the search window from the familiarization trials), a Person     Computer  Program   book, and a bird. The book and the bird were included to provide comparison points with the human and Figure 2. Total number of questions asked of each technological sources. The experimenter introduced information source. Dark sections indicate proportion each object by stating “This is a [object name]” and of questions that were causal in nature (i.e., “how” or then asked 7 questions in the form of “Can this one “why” questions). Error bars indicate SE. ____?” The questions addressed biological processes (“eat”), perception (“see things”), cognition (“think”), Attribution Trials Although four children did not emotion (“feel happy”), social awareness (“tell how ask questions, all 30 participants successfully you feel”), intentionality (“want to help you”) and completed the familiarization and attribution trials. pedagogical capacity (“teach you something”). Thus, we included every child tested in our analyses. For these trials, we calculated a score of 0-7 for the total number of characteristics children attributed to Results each object. We analyzed these data using a repeated- All children correctly identified the correct answers measures ANOVA with Object-type as a within- in the familiarization trials and no child was able to subjects factor. This test revealed a significant main initially name the animals correctly, confirming that the effect of Object-type, F(3,87) = 68.32, p < .001, ƞp2 = animals were unfamiliar to all of the children. .703. Bonferroni-corrected post-hoc analyses revealed Question Trials There were a total of 215 questions that significantly more attributions were made to the asked by 26 participants. (Four children did not ask human than to the bird, computer program, or book (ps questions of either informant.) For our initial analysis, < .001). The bird represented an intermediate level of we calculated the combined number of questions over attributions, which was significantly lower than the the two trials asked of each informant. Preliminary human, but still significantly greater than the computer analyses showed no gender differences nor effects of program or book (ps < .05). The number of attributions the order in which the informants or animals were to the computer program or book did not significantly introduced so these variables were excluded from differ, p = 1.00. further analysis. A paired samples t-test showed no Although these differences are important, the difference between the number of questions children attribution task included some characteristics that were asked of the human (M = 4.07, SD = 3.45) and the not essential to our goal of evaluating human versus computer program (M = 3.47, SD = 2.54), t(29) = 1.57, technological informants (e.g., questions about physical p = .124. or perceptual capabilities – questions that primarily In order to examine potential differences in the ensured that children were paying attention and knew content of the questions children directed to the human how human and technological informants differ). Thus, or technological informant, we calculated the we conducted a second analysis focused more narrowly proportion of questions each child asked of each source on attributions about thinking and teaching by that used the terms “why” or “how.” (The two “how” combining responses to the cognition and pedagogical questions related to the animal’s sleep and movement capacity items to create a more focused composite patterns, and were causal in nature.) We found that 38% score. We analyzed this data as we did the full set of of the questions directed to the human and 44% of the attributions, again finding a significant main effect for questions directed to the computer sought causal Object-type F(1,29) = 5.12, p < .05, ƞp2 = .15. Post-hoc explanations, yielding no significant difference in the analyses revealed a pattern that differed slightly from 410 our findings including all attributions. As in our initial information sources is important because children today analysis, significantly more attributions were made to learn from both sources, but unlike human sources, the human (M = 1.87) than to the bird, computer technological devices do not have the beliefs, program, or book, p < .001, but the bird, computer intentions, and behavior patterns that underlie human program, and book did not differ from each other (Ms pedagogical behaviors. between .57 and .67). Our data reveal that children’s question-asking behaviors were similar in quality and quantity for both 7   humans and computers. Children asked many questions, #  of  Characteristics   6   and the proportion of causal questions asked did not 5   vary between the two informants. One interpretation of these data is that children were more focused on the 4   subject of their questions (i.e., the unfamiliar animals) 3   than on the nature of the information source. They may 2   also view both information sources as equally capable of answering both causal and non-causal questions 1   about animals, although perhaps children would have 0   made a greater distinction between information sources Person   Bird   Computer   Book   if the questions were in another domain, such as moral program   reasoning (see Danovitch & Keil, 2008). Another Figure 2. Mean number of characteristics attributed to potential interpretation is that children of this age did each object in the attribution task. (Error bars indicate not understand the differences between the informants SE.) and therefore did not tailor their questions to them; however, their responses on the attribution task suggest Relationship between questions asked and that this is not the case. attributions We performed a correlational analysis to In the attribution task, we measured children’s investigate whether the content or number of children’s intuitions about the biological and psychological questions were related to their attributions of biological characteristics of human and technological informants. and psychological characteristics to the computer Despite treating human and technological information program. We found no significant correlations between sources very similarly in terms of their question-asking children’s rate of asking questions of each informant or behaviors, children’s attributions of characteristics to the proportion of causal questions asked and the human and technological sources were quite different. number of characteristics attributed to the computer Although children asked the same kinds of questions of program. There was also no relationship between the the person and the Internet search engine when allowed number or nature of the questions asked and children’s to do so, they did not explicitly state that the Internet attribution of the capacity to think and teach to the search engine had characteristics that would allow it to computer program. Thus, children’s questions appeared be an effective teacher (such as the capacity for to be largely independent of the characteristics that thought, intention, and pedagogy). Thus, children’s children attributed to technological informants. explicit understanding of computers as a non- pedagogical entity appeared to have no meaningful relationship to their question-asking behaviors. That Discussion said, relatively few children in our sample attributed We examined what kinds of information 4- and 5- pedagogical characteristics to the computer program. year-old children seek from a human or a technological This stands in contrast to the adult intuition that informant (i.e., an Internet search program), and how computer programs, and particularly Internet search their information seeking behavior relates to their engines, can be effective means of obtaining attributions of psychological and biological information and learning new concepts (Zickuhr, 2010). characteristics to technological informants. Our Thus, additional research is needed to investigate methods provided children with an opportunity to whether the pattern of responses we observed persists utilize one of their most powerful and natural learning over development, or whether it is unique to young mechanisms: asking questions. We presented children children who have relatively limited experience using with unfamiliar animals, knowing that these effectively technology to find information on their own. provoked questions in prior studies (e.g., Greif et al., These findings also raise important questions for 2006). Critically, we sought to evaluate whether future research regarding the broader consequences of children’s questions to a human versus a technological obtaining information from technological informants. informant varied in terms of volume and the proportion Why do children ask similar questions of information of causal questions. The contrast between these two sources that are teachers and non-teachers? Do children 411 co-opt question-asking behaviors that are usually animals and artifacts? Domain-specific requests for targeted at pedagogical sources, or do they treat information. Psychological Science, 17(6), 455-459. technological informants differently from other Koenig, M. A., Clément, F., & Harris, P. L. (2004). information sources? It would also be informative to Trust in testimony: Children’s use of true and false examine how children reconcile their implicit behavior statements. Psychological Science, 15, 694–698. toward technological informants (e.g., asking them Lutz, D. J., & Keil, F. C. (2002). Early understanding of questions of the same nature that they ask of humans) the division of cognitive labor. Child Development, and their explicit beliefs about technology, and whether 73, 1073–1084. this changes over the course of development. Mikropoulos, T. A., Misailidi, P., & Bonoti, F. (2003). In conclusion, our current findings represent an Attributing human properties to computer artifacts: important first step toward understanding how growing developmental changes in children's understanding of up surrounded by information technology affects the animate-inanimate distinction. Psychology: The children’s curiosity and exploration of information. As Journal of the Hellenic Psychological Society, 10, the devices that search the Internet become more 53-64. readily available and children encounter them at Mills, C. M., Legare, C. H., Bills, M., & Mejias, C. younger ages, it is essential that we understand (2010). Preschoolers use questions as a tool to children’s assumptions about the capacities and acquire knowledge from different sources. Journal of limitations of technological informants and use this Cognition and Development, 11(4), 533-560. understanding to inform the ways in which children Mills, C. M., & Keil, F. C. (2004). Knowing the limits interact with and learn from technology. of one’s understanding: The development of an awareness of an illusion of explanatory depth. Acknowledgements Journal of Experimental Child Psychology, 87(1), 1- This research was supported in part by funding from the 32. National Science Foundation, CAREER grant DRL- Richler, J. (2015) What children think of the Internet 1149116 to PS. Thank you to Jacob Messmer, Kayla and Why it Matters. Retrieved from: Renner, and Gretchen Santana for their assistance with http://parenting.blogs.nytimes.com//2015/02/16/what data collection. We would also like to thank the staff, -children-think-of-the-internet-and-why-it-matters/ children, and parents at St. Paul School and Keneseth Scaife, M., & Van Duuren, M.A. (1995). Do computers Israel Preschool in Louisville, KY. have brains? What children believe about intelligent artifacts. British Journal of Developmental References Psychology, 13, 367–377. Subrahmanyam, K., Gelman, R., & Lafosse, A. (2002). Bauerlein, M. (2008). The Dumbest Generation: How Animate and other separably moveable things. In G. the digital age stupefies young Americans and Humphreys (Ed.), Category-specificity in Brain and jeopardizes our future (or, don’t trust anyone Mind (pp. 341–371). London: Psychology Press. under 30). New York, NY: Penguin. Ward, A. F. (2013) One with the cloud: Why people Birch, S. A., Vauthier, S. A., & Bloom, P. (2008). mistake the Internet’s knowledge for their own Three- and four-year-olds spontaneously use others’ (Unpublished doctoral dissertation). Harvard past performance to guide their learning. Cognition, University, Cambridge, MA. 107, 1018–1034. Yan, Z., (2005). Age differences in children’s Carr, N. (2010). The shallows: What the Internet is understanding of the complexity of the Internet. doing to our brains. New York, NY: W.W. Norton. Journal of Applied Developmental Psychology, 26, Danovitch, J. H., & Keil, F. (2008). Young Humeans: 385-396. the role of emotions in children's evaluation of Yan, Z., (2006). What influences children’s and moral reasoning abilities. Developmental Science, adolescents’ understanding of the complexity of the 11(1), 33-39 internet? Developmental Psychology, 42, 418-428. Fisher, M., Goddu, M. K., & Keil, F. C. (2015). Yan, Z., (2009). Limited knowledge and limited Searching for explanations: How the Internet resources: Children’s and adolescents’ understanding inflates estimates of internal knowledge. Journal of of the internet. Journal of Applied Developmental Experimental Psychology: General. Psychology, 30, 103-115. Frazier, B. N., Gelman, S. A., & Wellman, H. M. Zickuhr, K. (2010). Generations 2010. Retrieved from (2009). Preschoolers’ search for explanatory Pew Internet Research website: information within adult–child conversation. Child http://www.pewinternet.org/files/old- Development, 80(6), 1592-1611. media/Files/Reports/2010/PIP_Generations_and_Tec Greif, M. L., Nelson, D. G. K., Keil, F. C., & Gutierrez, h10.pd F. (2006). What do children want to know about 412