Max Bulger PSY28: Review Paper

Food and Cognitive Recognition
Perception and Parsing Sensory Data

2 INTRODUCTION One of the most impressive aspects of human cognition is our ability to process sensory information and categorize objects in the world around us almost instantly. With a single glance, we can parse between furniture, aspects of nature, other humans, machines and endless other variations of artifact. One of the most complex and oft overlooked aspects of cognitive identification is our ability to classify and analyze food objects. The first study examined, Shutts et al. (2009), explores the possibility of an innate cognitive ability to identify food vs. non-food objects. Following the theory of core knowledge laid out by Spelke & Kinzler (2007), Shutts et al. (2009) specifically sought to discover whether or not humans possessed a domain for food within core knowledge. That is, are humans born with a cognitive system that helps us determine what is and is not food in the world around us, or is food identification an ability learned during childhood development? While Shutts’ research is inconclusive, it exposes the potential for a wealth of future research about the mental processes involved with interpreting food objects. The second study, Morrot et al. (2001), targets the connection between olfactory information processing and other cognitive processes. Specifically, the study seeks to discover how human interpretation of smell is impacted by visual information and the restraints of lexical output. By conducting a series of wine tasting trials and analyzing the semantic and lexical output by participants, Morrot concludes that sight and lexical output can create a sensory illusion that dramatically interrupts and alters human ability to interpret olfactory data. While the experimenters do not address this fact, it is interesting to note that olfactory data is the only sensory information that does not

3 undergo any processing before entering the thalamus. As with the first study, this provides fascinating opportunity for further research. It is clear the link between cognition and how we interpret what we eat and drink is strong and could be far more complex than some brain scientists believe. Cognitively, how do we identify food? Are there innate systems of classification or do we learn how to recognize food? Does one sense take priority of others in classifying food vs. non-food artifacts? Do we intake all sensory information than classify, or is there an earlier initial cue that tips us off to a food object and helps us codify incoming sensory information? When we classify something as food, we collect novel data and sensory information in a way that is potentially different from how we perceive other objects. Taste, smell, texture, and shape are all interpreted in a unique fashion. There is a strong need for further research to explore the link between cognition and food.

4 3. Review: (a) Shutts, K., Condry, K.F., Santos, L. R. & Spelke, E. S. (2009). Core knowledge and its limits: The domain of food. Cognition, 112, 1: 120-140. (b) EXPERIMENT 1 i. QUESTIONS: The experimenters seek to explore the possibility of a domain for food under the theory of core knowledge. Recognition tasks are highlighted as the primary indicator of the presence of a food-specific domain. In the first experiment, the experimenters seek to establish a baseline by duplicating and slightly updating a study on rhesus monkeys done by Santos et al. (2003). Do rhesus monkeys have a core system of inherent knowledge that would enable them to exhibit different generalization patterns for food vs. non-food object recognition? In other words, do rhesus monkeys have an inherent ability to parse food objects from artifacts, or is that an ability learned in cognitive development? ii. METHOD: The method for the first experiment was designed to mimc Santos et al. (2003)’s methodology, with a single modification. An experimenter held artifacts in front of a rhesus monkey, and the monkey’s facial reactions and eye movements were recorded and analyzed in accordance with each object presented. The monkeys were initially shown 3 familiarization trials, in which the experimenter held an L-shaped object of a distinctive color, and tasted the object (put it in their mouth and made a sound). Next, the monkeys viewed 2 test trials. For these tests, the procedure was identical, except the objects were different: in the first test, the object was a different color but the same shape as the object from the familiarization trials, and, in the second test, the object was the same color but a different shape. Again, facial response and eye movements were recorded and analyzed.

5 iii. FINDINGS: Monkeys in the first experiment spent significantly longer looking at the two novel objects in the test trials. Additionally, the monkeys spent longer looking at objects of novel color and same shape than objects of novel shape and same color. This was measured by tracking eye and facial movements and noting when the monkeys looked away for more than 2s . iv. INTERPRETATION: The experimenters compared the findings to results of the experiment these trials were designed from, Santos et al. (2003). The results from that experiment, identical in all aspects except that objects were presented as tools, not as food objects, were similar. Monkeys noticed novel objects. A key difference between the results here and in Santos, as noted by the experimenters, is that the results of the 2003 study showed monkeys paying more attention to changed shape. The results of this experiment showed color change to be more salient. The author used this disparity to emphasize the importance of the context in which objects are presented (specifically, here, whether they are tasted or not). The experimenters also suggested these results were evidence for domain-specific learning and recognition differences for food vs. nonfood objects. However, the monkeys in this experiment had previous exposure to food and experience recognizing it. To further isolate and test for core knowledge about food recognition, the experimenters concluded they needed to test human infants. (c) EXPERIMENT 2 i. QUESTIONS: This experiment pursued the greater theme of the study, examining the possibility of a food-specific domain in human core knowledge by testing object recognition. Specifically, these trials were design to test whether infants could recognize

6 and differentiate between two food objects based exclusively on visual features available in static display. ii. METHOD: 9-month old infants were seated in a high chair with a small stage directly in front of them. Just as in Experiment 1, their facial reactions and eye movements were recorded on video as different artifact stimuli were presented. When their eyes moved away for 2s or more, or 60s elapsed, that set of stimuli ended (because it was assumed that is when the infant recognized the object and ceased to focus on it). The two objects presented were a small pumpkin resting atop a ginger root, the second object. In both the experimental and control, the infants were first presented with a habituation stimulus: a hand descended and came to rest on the pumpkin. This stimulus was presented repeatedly until the infants were deemed ‘habituated;’ as defined by a significant drop in recorded time until the infants looked away. After the habituation stimulus, half of the infants were subjected to a ‘relative motion’ trial and half to a ‘common motion’ trial. In the relative motion trial, the hand descended as previously, but this time grasped the pumpkin and lifted it upwards and away from the ginger root. Reactions were recorded. In the common motion trial, the hand descended and grabbed the pumpkin, but lifted both objects into the air (the pumpkin and ginger root were secretly connected in a way invisible to the infant). Reaction times were recorded. iii. FINDINGS: No significant time difference was observed between the common motion trial and the relative motion trial. iv. INTERPRETATION: Experiment 2 provides no evidence that infants have core knowledge pertaining specifically to food or food recognition. The experimenter’s hypothesis was that, if infants possessed a domain for food, they would perceive the

7 pumpkin and ginger root as distinctly different objects and be puzzled when they were lifted together. However, the results showed the infants could not parse between the pumpkin and ginger root. The experimenters offer the possibility that the objects were not clearly enough food objects and that was the source of error. EXPERIMENT 3 i. QUESTIONS: Experiment 3 was designed in line with the experimental questions of the entire set of studies in regards to a domain for food in infantile core knowledge. The specific concern of this set of trials was with an infants ability to detect features of a single food object while it was whole compared to while it was broken into two pieces. ii. METHOD: The method of Experiment 3 mirrored that of Experiment 2. In Experiment 3, the habituation trial consisted of a hand descending from above and resting on a single lemon. There were no other objects present in the display. The control trial consisted of the hand lifting the whole lemon up off of the display stage. The variable trial consisted of the hand lifting just the top half of the lemon off of the stage. No cut or break in the lemon was apparent prior to the lifting in the experimental trial. Facial reaction and eye motion was recorded for each trial with the same 2s/60s rule as Experiment 2. iii. FINDINGS: After the habituation trials, the infant took significantly longer to recognize the halved lemon than the whole lemon. iv. INTERPRETATION: While Experiment 3 did show successful differentiation, it does not provide evidence for food-related core knowledge. The experimenters highlight that the results were identical to previous studies done by other experimenters with nonfood artifacts, meaning these findings do not advance any understanding of a food-

8 specific cognition process. The text highlights a potential source of error: the objects were not presented in an eating context, so perhaps the infants were not aware the lemons were food. EXPERIMENT 4 i. QUESTIONS: The greater investigation of a domain for food in the core knowledge of infants remains as previously. The fourth iteration of experiment in this line of research sought to refine the third experiment by ensuring it was clear to infants the objects in each trial were food artifacts. Experiment 4 was designed to test whether infants could detect property changes in different food objects. ii. METHOD: Infants were exposed to habituating trials observing an experimenter eating green sugar out of a champagne glass. After the habituation trials, the infants were split into two groups and exposed to two different trials. One group observed the experimenter eating a novel food (orange juice) out of the same container (champagne glass). The second group observed the experimenter eating the same food (green sugar) out of a novel container (a bowl). Recognition times were recorded through facial and eye movements, as in prior experiments. iii. FINDINGS: Infants took longer to recognize the objects in both experimental conditions. There was no significant different between recognition times for the two experimental trials (novel food and novel container had the same average recognition times). iv. INTERPRETATION: Experiment 4 suggests that infants have the ability to track changes in food properties, as displayed by longer processing times with novel food

9 objects or containers. However, it is unclear if this ability is food-specific, because each experimental trial only tested one property: novel food or novel container. EXPERIMENT 5 i. QUESTIONS: The question of core knowledge in infants pertaining to food remains. The experimenter specifically sought to see if infants could parse between changes in the container vs. changes in food object. ii. METHOD: The method for Experiment 5 mirrored that of Experiment 4, with the inclusion of more experimental conditions and use of nonsolid foods. Experimental trials included an experimenter eating a novel nonsolid food from a novel container, a habituated familiar nonsolid food from a novel container, a habituated familiar nonsolid food from a habituated familiar container, and a novel nonsolid food from a familiar container. Recognition response times were recorded using video recordings of eye movement. iii. FINDINGS: There was no significant correlation pattern or differences in recognition times amongst subjects. iv. INTERPRETATION: Experiment 5 provides further evidence against the possibility of a domain for food inherent in infantile cognition. If infants privileged color and texture when cognitively classifying food objects, they would have shown differing times with partially novel and entirely novel recognition tasks. Because all times were even, Experiment 5 does not support the experimenter hypothesis. EXPERIMENT 6

10 i. QUESTIONS: The questions behind the design of Experiment 6 are identical to those of Experiment 5. The only difference is Experiment 6 investigated with solid food objects, as opposed to nonsolid foods. ii. METHOD: The exact same habituation and experimental trials were utilized in Experiment 6 as Experiment 5. The only difference was the substitution of solid foods for nonsolid foods. iii. FINDINGS: There was no significant correlation pattern or differences in recognition times amongst subjects. iv. INTERPRETATION: Just as in Experiment 5, Experiment 6 provides no support to the hypothesis that core knowledge contains a domain for food. At this point, the experimenters acknowledged growing doubt for the viability of their original hypothesis. EXPERIMENT 7 i. QUESTIONS: The larger questions at hand in Experiment 7 are the same for Experiments 1-6. Experiment 7 specifically tested for the same food object and container property change recognition as Experiments 5 and 6, with the single difference being in presentation of objects. ii. METHOD: Experiment 7 used the exact same methods as Experiment 5 and 6. Infants were habituated to one food and container, and then exposed to novel foods and containers in multiple patterns. The single change between Experiment 6 and 7 was in how the objects were presented. The experimenter displayed each object as a rattle by picking it up, shaking it twice, and placing it back down. The exact same containers and food objects from Experiment 6 were used.

11 iii. FINDINGS: There was no significant correlation patter or differences in recognition times across all experimental conditions. No significant results. iv. INTERPRETATION: Infants do not seem to have a distinct core system of knowledge for food recognition. They fail to recognize changes in food objects and properties of such objects in the same patterns that human children, adults and adult monkeys display. Infants seem to remember properties of food substances, as exhibited by differences in habituated times vs. novel times, but they do not use this information to cognitively track changes or recognize novel objects. Because of this, there is no evidence of a food-specific domain.

4. Review: (a) Morrot, G., Brochet, F. & Dubourdie, D. (2001). The Color of Odors. Brain and Language, 79, 309-320. (b) EXPERIMENT 1 i. QUESTIONS: Morrot et al. (2001) sought to examine the cognitive processes related to olfactory information processing, especially in regards to visual sensory input and semantic translation. How do humans process olfactory information on a cognitive level? How does vision impact our sense of smell? How does our cognitive understanding of what we are smelling change as we try to codify it with a standard lexicon? ii. METHOD: 54 undergraduates from the Faculty of Oenology at University of Bordeaux were selected to participate in a wine tasting. In the control trial, students were presented with a glass of white Bordeaux wine (W) and a glass of red Bordeaux wine (R). Participants were not given any information about the wine. They were then given a list of terms to assign to each wine, and also offered the option of making up their own descriptors, and asked to describe the wine.

12 One week later, the same undergraduates returned for the experimental trial. The participants were presented with two glasses of wine in the exact same format as the control. One glass contained W from the first trial. The other glass contained wine W, dyed red with a completely odorless coloring agent (RW). Students were not told anything about the wines. They were provided with the same list of terms from the first trial, as well as a list of any descriptors they had created in the first trial. Students were not offered the chance to create their own terms in the second trial. They were asked to complete the same written description task. Lexical analyses were conducted using the computer program ALCESTE (for more information about ALCESTE, see Reinert, 1986). iii. FINDINGS: Participants used the same set of terms to describe RW in the second trial as they did to describe RW in the second trial. Additionally, participants eliminated terms used to describe W in the first trial from descriptions of RW in the second trial. Participants used the same set of terms to describe W in both trials. iv. INTERPRETATION: Visual data clearly clouded participants’ cognitive analysis of olfactory flavor information. The cognitive process of recognizing wine is more pattern recognition than sensory analysis. Semantic analysis of the lexical information collected about participant descriptors and how they were applied reveals that the only common theme for grouping terms is color association. There were different variations and subthemes among smaller groups, but grouping based on color characteristics was the only common method of descriptor grouping shared by all participants. Every other pattern differed between participants. This led experimenters to identify a sensory illusion and

13 conclude that visual information about color interrupts the cognitive ability to interpret olfactory flavor.

14 REFERENCES Reinert, M. (1986). Un logiciel d’analyse leciale: ALCESTE. Les Cahiers de l’Analyse des Donnees, XI, 471-484. Santos, L.R., Miller, C. T. & Hauser, M.D. (2003). Representing tools: How two nonhuman primate species distinguish between the functionally relevant and irrelevant features of a tool. Animal Cognition, 6, 269-281. Spelke, E. S. & Kinzler, K. D. (2007). Core knowledge. Development Science, 10:1, 8996.