People's motivational states-their wishes and preferences-influence their processing of visual stimuli. In 5 studies, participants shown an ambiguous figure (e.g., one that could be seen either as the letter B or the number 13) tended to report seeing the interpretation that assigned them to outcomes they favored. This finding was affirmed by unobtrusive and implicit measures of perception (e.g., eye tracking, lexical decision tasks) and by experimental procedures demonstrating that participants were aware only of the single (usually favored) interpretation they saw at the time they viewed the stimulus. These studies suggest that the impact of motivation on information processing extends down into preconscious processing of stimuli in the visual environment and thus guides what the visual system presents to conscious awareness. Keywords: motivation, visual perception, motivated reasoning, New Look, ambiguous figures The world that people know is the one they take in through their senses. This is the world they react to-the one their conscious thoughts, feelings, and actions are predicated on. People act on the presumption that the world they are consciously aware of is a comprehensive and accurate representation of the environment that exactly copies the outside world as it truly is. Decades of research in psychology, however, tend to undermine the assumption that what people see or hear is an exact replica of what is out in the world, in two different ways. First, perception is selective. People are not aware of everything that is going on around them. Consider, for example, recent studies of attentional blindness. Of undergraduates asked to monitor how many times people in a videotape pass a basketball among themselves, 40% failed to see the woman in a gorilla suit saunter into the middle of the group, turn to the camera, beat her chest, and then walk out Moreover, perception is malleable. It is responsive to top-down influences that flow from the perceiver's cognitive and psychological states or from environments But a substantial volume of psychological research reveals that top-down influences also inform perception. For example, context matters. Prior exposure to images of animals or people biases what people see when they view classic ambiguous figures, such as the rat-man and old woman-young woman figures so often featured in introductory psychology textbooks In the current article, we explore one possible top-down influence on perception that has been shown to have a profound and ubiquitous impact in other arenas of social cognition. That influence is the perceiver's motivational states-more specifically, the motivation to think of one's self and one's prospects in a favorable way, to believe that one will achieve positive outcomes while 2006, Vol. 91, No. 4, 612-625 0022-3514/06/$12.00 DOI: 10.1037 612 being able to avoid aversive ones, and to enhance self-worth and esteem. This motivation in the psychological literature has several names, such as motivated reasoning, self-affirmation, wishful thinking, and defensive processing, and has been shown to have a widespread influence in shaping how people think about their world, that is, how they interpret information of which they are consciously aware. This motive has been shown to influence such higher order tasks as judging other people, evaluating the self, predicting the future, and making sense of the past (for reviews, see In the studies that follow, we examine the scope of motivated reasoning to see if it crosses the boundary between how people think about their outside world and how they perceive it. Certainly, motivated reasoning influences conscious, deliberate, and effortful judgments, but we ask if it can constrain what information reaches consciousness in the first place. Does the impact of motivated reasoning or wishful thinking, more specifically, extend down to preconscious processing of visual information? We test, in essence, whether people literally are prone to see what they want to see. The Impact of Motivational States There exist some indirect hints that the motives underlying wishful thinking have an impact on visual perception. Recent work focusing on more biologically oriented motivational states shows that they influence the perception of visual stimuli. For example, But would a drive toward wishful thinking similarly influence perception? In a sense, this question is a revisiting and a reopening of one of the focal issues of the New Look approach to perception that arose in psychology during the 1940s and 1950s These initial demonstrations of motivational influences on perception were met with much enthusiasm, which was then followed by withering criticism. To be sure, much of what the New Look theorists proposed has lasted through today and informs contemporary cognitive and perceptual psychology in fundamental ways. Psychologists uniformly agree with the New Look tenet that much of cognition happens nonconsciously, that is, outside a person's awareness, monitoring, or control However, the specific New Look assertion that motivational states influence perception did not achieve the same stature and longevity as these other insights. It, instead, ran aground in the 1950s on the rocky shoals of methodological difficulties and theoretical controversies As such, the influence of motivational states on perception was never firmly established. And as the 1950s closed the study of the relation between motivational states and perception, this pursuit fell by the wayside and ceased to have the major impact-if any at all-enjoyed by other insights from the New Look tradition Perception of Ambiguous Figures In the present research, we examined the impact of motivational states on perception by focusing on interpretations of ambiguous or reversible figures-visual stimuli, like the famous Necker cube, that people can interpret in two different ways but for which they tend to see only one interpretation at any given time In each of five studies, we told participants that they were about to be assigned to one of two experimental tasks, one being much more desirable than the other. We also told participants that a computer sitting in front of them was about to present them a stimulus that would indicate which task they were assigned to. In fact, in each study, the computer presented a figure that could be interpreted in two different ways: one way that would assign participants to their favored task and one that would assign them to the opposite. We expected that participants would tend to see the interpretation that assigned them to the outcome they favored. MOTIVATIONAL INFLUENCES ON VISUAL PERCEPTION Because our experimental stimuli, like much of the contents of our surroundings, lack clarity and contain multiple interpretations, potential interpretations of a visual stimulus can be likened to a hypothesis Just as expectancies and contexts can suggest a testable perceptual hypothesis, a preference or desire might privilege a favored interpretation or hypothesis over a disfavored one. Wishful thinking might shape the specific hypothesis that individuals test when given such ambiguous information. In particular, the perceiver might scan the visual stimulus in a biased manner, searching for features that match those of the desired animal rather than those that match an undesired one. The net effect of focusing on a hypothesis is that the perceiver tends to seek out information that would confirm it rather than disconfirm it Alternatively, a motivated preference might lower the threshold required for the visual system to decide it matches the favored interpretation. Other work in motivated reasoning has shown that information consistent with a favored conclusion is held to a lower standard of scrutiny than information consistent with an unwanted one The key to whatever process is at play is that it takes place preconsciously. People are not aware that they have selected one interpretation over another. Indeed, they are not even aware of the alternative interpretation. Whatever work the visual system has done to bias the interpretation that people see involves processes below the level of awareness. Overview of Studies Studies 1 and 2 demonstrated that participants tended to report seeing the interpretation of an ambiguous figure that fit with their wishes and preferences over one that did not. Studies 3 and 4 added implicit measures to ensure that participants truly saw the interpretation they reported rather than simply reporting the preferred interpretation. Study 5 added a procedural twist to affirm that participants saw only the interpretation they usually wanted to see as they viewed the stimulus and that it was not the case that they saw both interpretations and then only reported the favored one. In short, people tended to honestly see only that interpretation that was suggested, in part, by their motivational state. Study 1: Disambiguating an Ambiguous Method Participants. Participants were 88 undergraduates at Cornell University who earned extra credit in their psychology or human development courses for taking part in the study. Procedure. In what was advertised as a taste-testing experiment, an experimenter explained that participants would predict taste sensations for two beverages, consume only one beverage, and describe their actual taste sensation of that one beverage. On the table in front of participants sat the two beverages. The first was the desirable one: freshly squeezed orange juice. The second was the less desirable alternative: a gelatinous, chunky, green, foul-smelling, somewhat viscous concoction labeled as an "organic veggie smoothie." 1 The experimenter invited participants first to smell each beverage. Then, participants spent 3 min predicting what they might experience if asked to drink 8 ounces (about 240 ml) of each beverage to heighten the appeal of the orange juice and strengthen their disgust with the veggie smoothie. Participants were seated in front of a 15-in. G3 iBook. The experimenter then explained that a computer program would randomly select a beverage for the participant to consume. Specifically, the computer would select either a single letter or a single number from a set of 26 letters and 26 numbers. Roughly half of the participants, those in the number-desirable condition, were told that if the computer selected a number from the set, they would drink 8 ounces (about 240 ml) of orange juice, and if a letter was selected, they would drink 8 ounces (about 240 ml) of veggie smoothie. The remaining participants in the letter-desirable condition learned that a letter would result in their assignment to the orange juice and a number to the veggie smoothie. After inviting the participant to review these directions on a computer screen, the experimenter stepped away to ostensibly complete some paperwork. Participants focused on the center of the monitor on which was displayed a static fixation point. After 3 s, this fixation point was replaced with an ambiguous figure (1 in. in height, 1 in. in width) that could be interpreted as either the capital letter B or the number 13 (see After receiving an answer, the experimenter handed the participant a questionnaire to complete while she supposedly left to prepare the bever-1 Recipe available on request. BALCETIS AND DUNNING age. This questionnaire probed for suspicion of the purpose of the study, suspicion of the computer crash, and in a funneled manner queried participants to see if they realized the ambiguity in the figure shown before the computer crash. Results A priori, we established conditions for the inclusion of participants' data. Participants were excluded if they recognized the figure was ambiguous, were able to explain the purpose of the study in debriefing, or mentioned they wished to be assigned to what was considered by most participants to be the less desired task (i.e., consumption of veggie smoothie). Given these criteria, 15 people were excluded for recognizing the ambiguity in the figure when viewing the figure, 4 for explaining that we were interested in how their desires could influence the way they saw the figure, 3 for stating they hoped to consume the smoothie, and 3 simply refused to participate when they heard that they might be asked to consume the smoothie. This left data from 63 participants for analysis. Although a few participants indicated the computer crash was suspicious, none of these participants were able to describe the purpose of the study or the reason for the crash. Responses from those 63 participants were coded by means of the following method. Reports of the letter B were given a score of ϩ1, and reports of the number 13 a score of Ϫ1. Those who did not offer a response or indicated that nothing was shown before the crash received a score of 0. We then subjected these scores to an ordinal logistical regression analysis (the constrained range of the coding system made more usual statistical procedures less appropriate) to see if participants tended to see different interpretations of the ambiguous figure depending on which interpretation was more desirable. As expected, participants' desire to see either letters or numbers influenced their interpretation of the B-13 ambiguous figure, 2 (1, N ϭ 63) ϭ 23.92, p Ͻ .001. In particular, when hoping to see a letter, 72% (n ϭ 18) of participants reported seeing the capital letter B, whereas 0% reported seeing a 13. When hoping to see a number, 60.5% (n ϭ 23) reported seeing a 13 and 23.7% (n ϭ 9) reported seeing the B. Some people in each condition reported that in fact nothing was shown before the crash (28%, n ϭ 7, in the letter-favorable condition; 15.8%, n ϭ 6, in the number-favorable condition). Our specific prediction focuses on the responses of those who offered an interpretation of the figure. When excluding those responses from participants who reported that nothing was shown before the crash, participants' desire to see either letters or numbers influenced their interpretation of the B-13 ambiguous figure, 2 (1, N ϭ 50) ϭ 23.96, p Ͻ .001. Additionally, we can collapse across the specific character participants were motivated to see and look at just the reported interpretation for those participants who offered one. In fact, 82% (n ϭ 41) of participants reported the desired interpretation, 2 (1, N ϭ 50) ϭ 20.48, p Ͻ .001. In addition, including those people in the analyses who indicated that the figure was ambiguous does not change this pattern, as similar numbers of participants across both motivational conditions reported the ambiguity of the figure (n ϭ 8, when hoping to see letters; n ϭ 7, when hoping to see numbers). That is, we gave a score of 0 to those people who indicated the figure was ambiguous and again conducted an ordinal logistic regression. Still, participants' desire to see either a letter or a number influenced their interpretation of the ambiguous Discussion In sum, Study 1 provided evidence that people's motivational states can influence their interpretation of ambiguous objects in their environment. When faced with an ambiguous figure that could be interpreted as either a number or letter, the interpretation that reached consciousness and was reported tended to be the one that placed participants in a desirable circumstance rather than in an unwanted one. However, it is possible that the participants' responses did not reflect their true percept. Instead of reporting what they saw, they instead just offered a report that assigned them to the orange juice. Put simply, participants may have lied about what they saw. Although we suspect this is not the case, we conducted a follow-up to assess this counterexplanation. In a design similar to Study 1, 28 participants were either motivated to see letters or numbers to avoid the veggie smoothie but were then shown unambiguous figures of B or 13, rather than an ambiguous figure, during the computer assignment process. For half of the participants, a letter assigned them to the orange juice, whereas for the other half a number assigned them to the veggie smoothie. Crossed with this, half of the participants were shown a B and the other half were shown a 13, resulting in a 2 (desired character: letter or number) ϫ 2 (character shown: B or 13) factorial. The alternative account predicts that participants' reports of the figure shown to them would be influenced by which character was desired as well as what character was shown to them. However, inconsistent with that account, we found that what participants reported depended only on the character shown to them. In all conditions, 100% of participants (n ϭ 7 in every cell) reported the actual figure shown, regardless of what figure was shown to them and what participants were motivated to see. Study 2: Replication Study 2 was designed as a conceptual replication involving a different ambiguous figure and a different procedure. In addition, in Study 1, we noted that a small but notable minority of partici- MOTIVATIONAL INFLUENCES ON VISUAL PERCEPTION pants was able to spot the ambiguity of the figure we showed them. In Study 2, we used a figure whose ambiguity was more opaque and thus not as likely to be noticed by participants. Method Participants. Participants were 52 undergraduates at Cornell University who received extra credit in their psychology course for taking part. Procedure. Participants completed a task ostensibly about differences in predictions of and actual taste experiences. The experimenter explained that participants would be experiencing and describing different taste sensations. Participants would predict taste sensations for three food items but actually consume only one of them. First, participants predicted what each of the following items would taste like: a bottle of Aquafina water, a bag of Jelly Belly candies, and a bag of gelatinous and partially liquified canned beans. After participants predicted taste sensations of each item, participants were seated in front of a 17-in. iMac 64 desktop computer. Again, supposedly to eliminate bias from the selection process, a computer program would randomly assign the item participants would consume. The experimenter explained that participants would play a game, and their final score would determine what item was consumed. In this game, the computer displayed pictures of animals worth positive and negative points. On the top of their response sheet was a table listing every animal that could be selected and the specific number of points each animal was worth. For half of the participants, farm animals were worth positive points, whereas sea creatures were worth negative points. For the other half of participants, this was reversed. Black and white drawings of the full bodies, heads, and artistic renditions of animals were displayed in the rounds that preceded the final round. Although the computer would be keeping an ongoing tally of the points accumulated, participants recorded the animal shown to them, the points that animal was worth, and their ongoing score ostensibly to corroborate the computer program. If their score at the end of 15 cards was zero, participants would consume the water. If their score was positive, they would consume the candies, but if their score at the end was negative, participants would consume the canned beans. Although participants were told that the program randomly selected animals from a set of four farm animals and four sea animals, the program was actually rigged such that every participant experienced one of two sequences of animals and point tallies, depending on what category of animal was worth positive point values. As the game progressed, ongoing scores, predetermined and consistent across participants, fluctuated between positive and negative. However, the last three rounds brought increasingly negative point totals. That is, ongoing scores became ever more suggestive that participants would consume the canned beans. Ongoing scores at the end of the penultimate round were such that only one animal was worth enough positive points to be able to pull participants from the negative and bring a positive final score, thus avoiding the canned beans. For half of the participants, this animal was a horse; for the other half, it was a seal. The animal displayed during the final trial was in fact an ambiguous figure (2.75 in. wide, 3.75 in. tall) that could be interpreted as either the head of a horse or the full body of a seal (see After the game, participants completed a funneled debriefing that probed for suspicion of the purpose of the study, possible alternate interpretations of the figure, and asked if they had seen the figure before. Results Given the criteria we established a priori, 5 participants were excluded for articulating the purpose of the study and 4 for mathematical errors that precluded them from desiring the target animal. No one reported seeing both interpretations of the ambiguous figure. These omissions left data from 43 participants for analysis. We used the same type of coding scheme for interpretations as in the previous studies. Given the natural bias of this ambiguous figure was to see a horse, those who reported a horse received a score of ϩ1. Because the less common interpretation of the figure was as a seal, those who reported a seal received a score of Ϫ1. Using an ordinal logistic regression, we found that participants' interpretations depended on what category of animal was worth positive points, 2 (1, N ϭ 43) ϭ 6.89, p ϭ .009. When hoping to see a horse, 66.7% (n ϭ 14) of participants saw the figure as a horse, and 33.3% (n ϭ 7) saw a seal. However, this bias reversed when hoping to see a seal. Only 27.3% (n ϭ 6) of this group saw a horse, but 72.7% (n ϭ 16) reported a seal, 2 (1, N ϭ 23) ϭ 6.70, p ϭ .01. Discussion In sum, Study 2 replicated the findings of the first study with a different figure and experimental procedure. Participants tended to see the interpretation of the figure that they desired to see, rather than one they wished to avoid. In addition, no participant, either spontaneously or in debriefing, noted the ambiguous nature of the figure they saw. However, a reader can propose one counterexplanation for these findings, one that we decided to test in a control study. Given that the three rounds preceding the ambiguous figure included animals 616 BALCETIS AND DUNNING that brought participants' scores down, it is possible that participants' expectations about the next type of animal and not their desire predisposed them to see an animal worth positive points. That is, participants fell prey to a gambler's fallacy, assuming that a run of negative scores made positive-scoring animals more likely to appear next. To test this alternative explanation, we reran a version of Study 2, asking participants to follow along with the computer game and to record their points on a response sheet. However, we made clear to them that they would not be consuming any products after the game and that there would be no consequence for the final score they earned. Instead, they were to act as proofreaders, reading the directions thoroughly and evaluating the clarity of them. As was the case in Study 2, half of the participants encountered a game that made the horse the most valuable animal, whereas the other half were led to believe the seal was the most valuable animal. Thus, this group of participants, aware of the point structure and the progression of animals, would also be susceptible to the gambler's fallacy but would have little reason to be motivated to see the most valuable animal in the final round. In this control study, interpretations of the figure were not biased by what animals were most valuable. Those for whom farm animals would have been the most valuable were not more likely to see a horse than were those for whom sea animals would have been the most valuable, 2 (1, N ϭ 40) ϭ 0.11, p ϭ .74. When farm animals were the most valuable, 65% (n ϭ 13) of participants saw the figure as a horse, and 35% (n ϭ 7) saw it as a seal. When sea creatures were the most valuable, 70% (n ϭ 14) saw the figure as a horse, and 30% (n ϭ 6) saw it as a seal. The results of this study can be compared with those of Study 2 to suggest that reducing desire to see a particular animal can reduce the bias in interpretations. Because we are making comparisons across studies, it is necessary to use a Stouffer's Z test (see Study 3: Adding an Unobtrusive Measure Study 3 was designed to provide convergent evidence that the interpretations participants reported were, indeed, the sole interpretations that came to consciousness as they viewed the ambiguous stimulus. One can propose, instead, that participants saw both interpretations and then simply chose the one to tell the experimenter that placed them in a happier circumstance. One way to test whether participants saw only one versus both interpretations is to collect more unobtrusive measures that participants would not suspect were designed to test which interpretation they had seen-if they knew the measure was being taken at all. As was the case in the previous studies, we asked participants to provide a verbal or written report of whether they had seen a horse or a seal after being shown a figure that could be interpreted as either. However, in addition, we also measured participants' eye movements to see if they would give clues as to how participants had interpreted the figure. Recent evidence suggests that initial eye movements on presentation of a stimulus are not influenced by conscious processing Method Participants. Participants were 79 undergraduates at Cornell University completing the study in exchange for extra credit. Procedure. Participants came into the lab alone and were seated approximately 20 in. from a 21-in. Apple cinema-display monitor (17 in. viewable). As was the case in previous studies, participants completed a task ostensibly about differences in predictions of and actual taste experiences of Aquafina, orange juice, and veggie smoothie. After participants predicted taste sensations of each item, the experimenter explained that to eliminate bias from the selection process, a computer program would randomly assign the item they would consume on the basis of their score at the end of a game similar to the one used in Study 2. As described in the previous study, the computer displayed pictures of farm and sea animals counterbalanced between participants to be worth either positive or negative points. Participants kept a record of the animal shown to them, the points that the animal was worth, and their ongoing score, ostensibly to corroborate with the computer program. Participants were told that although the computer would be keeping an ongoing tally of the points accumulated, they would still categorize the animal as either a farm animal or sea creature by clicking on a box on the computer screen to advance the computer to the next animal. The program displayed each animal for 1,000 ms, followed by a 500-ms blank screen, and finally a request to categorize the figure, which remained on the screen until participants responded. On the extreme left side of the categorization screen was a box labeled "farm animal," and on the extreme right was a box labeled "sea creature." Participants were instructed to categorize the animals on the computer correctly to avoid point penalties. In addition to losing points for incorrect categorization, participants learned that a portion of their final score would be determined by the speed of their categorization; thus, they were advised to categorize animals as quickly as possible. Unbeknownst to the participants, a video camera was hidden approximately 15 in. behind the monitor and trained on participants' eyes. Thus, every time the categorization task appeared on the cinema-display monitor, we were able to capture participants' initial eye movements. As practice to familiarize them with the task of viewing and categorizing animals, participants categorized filler animals eight times. After this practice session, participants completed 15 trials, the last of which displayed the ambiguous figure. Thus, participants were well-acquainted with the three-step process to complete a single trial: (a) view the animal, (b) categorize the animal on the computer screen, and (c) record the animal and points on the written response sheet. We were interested in the way in which participants interpreted the ambiguous figure. Their interpretation was measured in two ways: the written self-report and participants' eye movements immediately on perceiving the categorization screen. Given that initial eye movements are not influenced by conscious processing We expected then that desire to see a particular animal would influence the way that the ambiguous figure was reported on the response sheet. Specifically, we expected that participants, hoping to drink orange juice, would see the most valuable animal. In addition, we expected that participants' eye movements would corroborate their self-reports such that initial saccades would be toward the box labeled as the most desired animal. MOTIVATIONAL INFLUENCES ON VISUAL PERCEPTION Coder reliability. A coder, blind to condition, hypotheses, and purpose of the study watched the videotaped eye movements and noted the initial direction of movement for half of the data set. For the other half of the data set, a second coder, blind to condition, coded the videotaped eye movements. A third coder, blind to condition, randomly selected 18 participants from the complete data set and noted the initial direction of eye movement. Eye movements recorded by this third coder then served as a measure of interrater reliability. Across 213 individual trials from the 18 randomly selected participants, the third coder and the original coder agreed in 92% of the cases. If there was disagreement, the direction of eye movement as indicated from the original coder was used in analyses. In addition, to assess the validity of our nonconscious measure of initial eye movement and to see whether eye movements corresponded to what participants later reported, we randomly selected 48 participants and coded their eye movements in response to the 10 unambiguous animals that preceded the ambiguous figure. Across 480 trials, initial eye movements went to the correct categorization box 86% of the time. Results Explicit reports. Using the same coding scheme as in the previous studies that used the horse-seal ambiguous figure, we again ran an ordinal logistic regression. As expected, desire facilitated the disambiguation of the figure, 2 (1, N ϭ 79) ϭ 5.62, p Ͻ .02. When hoping to see farm animals, 83.7% (n ϭ 36) of participants saw the figure as a horse, and 16.3% (n ϭ 7) saw a seal. However, the pattern changed when participants hoped to see sea creatures. That is, 58.3% (n ϭ 21) of this group saw a horse, 33.3% (n ϭ 12) reported a seal, and 8.3% (n ϭ 3) of participants did not indicate their interpretation. When looking only at the interpretations of those who offered one, it appears that desire influenced the disambiguation of the figure. Those who were motivated to see farm animals were more likely to report seeing a horse than were those who were motivated to see sea animals, 2 (1, N ϭ 76) ϭ 4.02, p Ͻ .05. Eye movements. We used the same coding scheme in analyzing the interpretations gathered from participants' eye movements. Again, those whose initial look was to the farm animal box received a score of 1, those who initially looked to the sea creature box received a score of Ϫ1, and those who looked down to their response sheet and not to either the farm animal or sea creature box received a score of 0. We conducted an ordinal logistic regression and found that desire facilitated the disambiguation of the figure, 2 (1, N ϭ 79) ϭ 10.24, p Ͻ .001. When hoping to see farm animals, 62.8% (n ϭ 27) of participants looked to the farm animal box, 14.0% (n ϭ 6) looked to the sea creature box, and 23.3% (n ϭ 10) looked down to their score sheet. However, the pattern changed when participants hoped to see sea animals. That is, 30.6% (n ϭ 11) looked to the farm animal box, 41.7% (n ϭ 15) looked to the sea creature box, and 27.8% (n ϭ 10) looked down to their score sheet. When looking only at the interpretations of those who looked to either box, it appears that desire influenced the disambiguation of the figure. Those who were motivated to see farm animals were more likely to look to the farm animal box than were those who were motivated to see sea animals, 2 (1, N ϭ 59) ϭ 9.90, p ϭ .002. We should note that scores on our eye-tracking measure significantly correlated with the score participants received from their explicit reports (Spearman's ϭ .42, p Ͻ .001). Study 4: Converging Evidence from Lexical Decision Data Study 4 served as a conceptual replication of Study 3 but used a different type of indirect measure of perception. A good deal of research (e.g., Neely, 1991) suggests that a picture of an object serves as a prime for concepts associated with that object, even if people are not aware that they have seen the object (e.g., However, we also collected reaction time data to gain an additional measure of whether participants had specifically seen the interpretation they had reported-and only that interpretation. Just after viewing the figure, participants completed a lexical decision task (LDT) in which they were presented with letter strings and had to decide whether those letter strings formed English words. Each participant saw a word related to the concept of "horse" (e.g., cowboy) or "seal" (e.g., blubber). We predicted that participants would respond more quickly to a word in the LDT exercise when that word was related to the interpretation they preferred to see rather than to the opposite interpretation. If participants actually saw both interpretations, no such difference should be seen in participants' decision speed to words related to desired versus undesired interpretations. We also wanted to make sure that participants' interpretations of the ambiguous figure were indeed responsible for priming their reactions in the LDT, rather than an overall desire to see a farm animal or sea creature. Thus, as a control condition, roughly half of the participants responded to the LDT just before they saw the ambiguous figure rather than just afterward. If participants responded more quickly to desired-concept words to a greater degree after they viewed the ambiguous figure, that fact would suggest that the interpretation participants saw was the one influencing the speed of their lexical decisions. However, if just a desire to see one type of animal over the other is enough to prime performance in the LDT, then desired-concept words should be facilitated in both before and after conditions to an equal degree. This design also allowed us to investigate one mechanism by which participants' perceptions were influenced. Collecting LDT reaction times just before participants viewed the ambiguous figure allowed us to gauge whether people's preferences suggested a perceptual set Method Participants. Participants were 166 undergraduates at Cornell University who received extra credit in their psychology courses for taking part. Procedure. Participants came into the lab in groups of 2 to 4 to complete a task ostensibly about differences in internal and external evaluations of vocal abilities. The experimenter explained that approximately 75% of participants would evaluate various aspects of a person's vocal performance, whereas the remaining 25% would be asked to perform a tune as if in a karaoke bar. The experimenter clarified that these 618 BALCETIS AND DUNNING percentages meant that approximately 1 person in each session would be the singer and subject of evaluation, whereas the remaining people would be observers. After performing a tune, singers would evaluate their own vocal abilities on rhythmic ability, skill, and general appeal. The experimenter explained that these scores would be corroborated against those provided by the observers on the same dimensions. At this point, participants were shown a 60-s video clip ostensibly of past participants and observers completing the performance evaluation portion of the experiment to heighten anxiety about the potential assignment to the singer role. In this video, a stocky Italian man in his early 20s held a microphone while singing and dancing along to Gloria Gaynor's 1979 rendition of "I Will Survive." Participants were seated approximately 24 -26 in. from a 17-in. iMac G4 or a 17-in. eMac desktop computer. As was the case in previous experiments, the experimenter explained that to eliminate bias from the selection process, a computer program would randomly assign participants to either the role of singer or observer. Participants played the same animal game as described in Study 3, ostensibly to determine whether they danced or observed. Again, participants kept a record of the animal shown to them, the points that the animal was worth, and their ongoing score, ostensibly to corroborate with the computer program. Additionally, participants categorized the animal as either a farm animal or sea creature on the computer. Finally, participants completed a number of LDTs during the animal categorization task, supposedly meant to impair their ability to categorize the animals. That is, participants categorized strings of letters as words or nonwords. In a go/no-go paradigm, participants hit the space bar if the string of letters was a word and did nothing if the string of letters was not a word. All strings of letters disappeared from the screen if no key was hit within 2,000 ms. Participants randomly assigned to the control condition completed the LDT at the beginning of each trial, that is, before seeing each animal. Participants randomly assigned to the experimental condition completed the LDT at the end of each trial, after seeing each animal but before categorizing it on the computer or recording it on their response sheet. Participants completed between one and three lexical decisions during each trial for the first 12 trials. In the last round, participants responded to three strings of letters. In this last trial, all participants responded to one word related to farm animals, one related to sea animals, and one nonword, the order of which were counterbalanced between subjects. Although for each participant only a single farm-and sea-relevant word was included in the last trial, the particular word selected was counterbalanced between subjects. Specifically, there were four words related to farm animals (cowboy, saddle, stallion, pasture), four words related to sea animals (blubber, flipper, ocean, whale), and four nonwords (blevre, yaver, dreas, puli) that were varied between subjects. That is, a participant would react to a single word from each of these sets. Again, the ongoing score at the end of the penultimate round were such that only one animal was worth enough positive points to produce an assignment to the observer role. For roughly half of the participants, the only animal capable of this was a horse, whereas for the other half, it was a seal. The last animal displayed was again the horse-seal ambiguous figure. We presumed that participants, going into the final trial with a negative score, would be hoping to see the animal worth the greatest number of positive points. We expected then that desire to see a particular animal would influence the way that the ambiguous figure was interpreted. Additionally, we expected that the desire to see a particular set of animals would influence the speed at which the target words was categorized, but only after participants had viewed the ambiguous figure. In particular, we expected that the control group that completed the LDTs before seeing the ambiguous figure would be equally likely to categorize the horse-relevant fragments and seal-relevant fragments as words. However, we expected that the experimental condition that completed the LDTs after having seen the ambiguous figure and interpreted it as the desired animal would be faster to categorize words related to the desired animal type. Specifically, those participants in the experimental condition for whom farm animals were worth positive points were expected to categorize the farm-relevant words faster than sea-relevant words.

Abstract Amongst philosophers and cognitive scientists, modularity remains a popular choice for an architecture of the human mind, primarily because of the supposed explanatory value of this approach. Modular architectures can vary both with respect to the strength of the notion of modularity and the scope of the modularity of mind. We propose a dilemma for modular architectures, no matter how these architectures vary along these two dimensions. First, if a modular architecture commits to the informational encapsulation of modules, as it is the case for modularity theories of perception, then modules are on this account impenetrable. However, we argue that there are genuine cases of the cognitive penetrability of perception and that these cases challenge any strong, encapsulated modular architecture of perception. Second, many recent massive modularity theories weaken the strength of the notion of module, while broadening the scope of modularity. These theories do not require any robust informational encapsulation, and thus avoid the incompatibility with cognitive penetrability. However, the weakened commitment to informational encapsulation greatly weakens the explanatory force of the theory and, ultimately, is conceptually at odds with the core of modularity. Amongst philosophers and cognitive scientists, modularity remains a popular choice for an architecture of the human mind. Jerry There are compelling theoretical and empirical motivations for this approach. Theoretically, modularity nicely accommodates adaptationist and other evolutionary explanations of mental phenomena. It also provides materials for a simple explanation of important empirical data, including a wide range of behavioural dissociations, as well as the speed and robustness of processing enjoyed by the human mind. Most broadly, modularity provides an intuitive framework for characterizing the relations between brain structures and particular perceptual and cognitive functions. Although it is sometimes misrepresented as doing precisely this, Fodor's pioneering discussion of the concept did not involve a definition of 'module'. Fodor maintains that "The notion of modularity ought to admit of degrees" (Fodor 1983: 37), and that "if a psychological system has most of the modularity properties, then it is very likely to have all of them" (Fodor 1983: 137). Importantly, Fodor claimed only that input systems are modular. His primary subject matter was perceptual systems, but he also made the case for systems devoted to low-level linguistic decoding. Higher-level conceptual or cognitive systems, then, are not modular on Fodor's general architecture. Commitments with respect to Fodor's original analysis of modularity vary. Several modularity theorists take domain-specificity to be definitive of modularity (Coltheart Our suggestion is that informational encapsulation is essential to a distinctive, nontrivial modularity theory. As it will be understood here, if a module m is informationally encapsulated then m cannot, during the course of its processing, access or compute over information found in other components of the overall system. As such, an encapsulated module m is impenetrable with respect to the other components of the system, since the processing of m is insensitive to (and so does not compute over) the information available elsewhere in the system In this respect, the modularity theorist faces a dilemma that hinges on the commitment to informational encapsulation. On the one hand, a commitment to informational encapsulation, as made by modularity theories of perception, is inconsistent with the cognitive penetration of perceptual experience. And, we argue, there are genuine cases of the cognitive penetrability of perception. On the other hand, 1 In at least two places, Fodor himself explicitly states that "informational encapsulation is an essential property of modular systems" (Fodor 1985: 3; see also 1983: 71). Elsewhere, however, he is less clear on his commitment regarding the same claim. 4 as recent modularity theorists have done, one might weaken the notion of module so as not to require informational encapsulation. The result, however, is an account that undermines one of the central motivations for modular architectures and, more fundamentally, that may not be consistent with the conceptual core of the very notion of modularity. The first horn challenges strong, encapsulated modularity: any modularity theory that includes a commitment to informational encapsulation. The second horn challenges massive modularity, which broadens the scope of modularity but weakens the notion of modularity so as not to require informational encapsulation. Either way, the modular approach to the study of cognitive architecture is significantly challenged. I. Informationally encapsulated modules: Cognitive penetrability and the challenge for encapsulated modularity Both encapsulated and unencapsulated modularity theorists take perceptual systems to be modular. If perceptual modules are informationally encapsulated, then at the very least, they are not penetrable by the information or processing of higher-level cognitive systems. Most theorists seem to take the concepts 'informational encapsulation' and 'cognitive impenetrability' to be co-extensive, if not equivalent-Fodor in fact originally argued for the encapsulation of modular input systems by arguing against claims about the cognitive penetrability of those systems (Fodor 1983: 73-86). The following discussion requires only the assumption that informational encapsulation of perceptual modules entails cognitive impenetrability. On this account, then, perceptual processing is not influenced by cognitive states like belief or desire. Evidence of this influence-that is, of the cognitive penetration of 5 perception-thus threatens any modularity theory that includes a commitment to informational encapsulation of perceptual modules. It will be useful here to offer some clarifications. First, distinguish perceptual experience from higher-level cognitive and affective states and processes like belief, judgement, desire, emotion, and so on. Perceptual experience, whatever else one says about it, is characterized by phenomenal character or content and depends non-trivially on one or more sensory organ. Philosophers debate how to draw the line between perception and cognition. The only point that need be granted here is that there are clear cases of perceptual states and clear cases of cognitive states. So there are visual experiences, auditory experiences, olfactory experiences, and so on; and these can be distinguished from states like belief and processes like decision making. Second, distinguish the cognitive penetration of perceptual experience from the cognitive penetration of perceptual processing. The former concerns some difference in the phenomenal content or character of a perceptual experience, where this difference depends non-trivially upon some cognitive state or processing in the system. The latter only concerns some cognitive effect on perception at the level of processing. The fact that perceptual processing at some stage is cognitively penetrated does not, by itself, entail the cognitive penetration of experience. Experience may depend on a wider class of processing and, in principle, the cognitive influences on perceptual processing (at some particular stage or other) may not ultimately influence conscious experience. Moreover, some aspects of perceptual processing may not give rise to a conscious experience but rather, for example, to the sub-personal guidance of motor performance. However, cognitive penetration of perceptual experience does entail cognitive penetration of perceptual processing at some level. There is much to be said here. The only assumption we need regarding the relation between perceptual experience and perceptual processing is this. Whether one takes experience to be identified with, 6 constituted by, supervenient upon, or the output of perceptual processes, a difference in perceptual experience implies a difference in perceptual process. , then these claims all concern descriptive vision. Accordingly, an effect on perceptual experience is an effect on descriptive vision which-according to a dominant theory in cognitive neuroscience-is plausibly an effect on processing in the neural pathway known the ventral stream (see 7 penetrated elsewhere such that the resulting perceptual experience is causally dependent upon cognition 3 Although there is ample evidence for "top-down" processing in the brain-where information is exchanged between various areas of the cortex, including those areas believed to process higher-level or conceptual information-current neuroscience lacks an uncontroversial mapping from conceptual mental states (like belief) onto brain structures. And some such mapping would be necessary for neuroscience to provide a verdict on the actuality of cognitive penetration. 4 Consequently, empirical evidence for cognitive penetration must be obtained at the behavioural or psychological level, rather than merely the neurological level. Predictably, there are a number of possible alternative interpretations of this data, and so the inference structure is abductive. Critics of cognitive penetrability appeal to these alternative interpretations as better explaining alleged cases of cognitive penetration. We identify four such general skeptical strategies. With these strategies in hand, a working definition of cognitive penetration can be devised, in hopes of isolating the target phenomenon in a way agreeable to both sides of the debate.

Israel has known a very high inflation rate and two currency changes in the past five years. Coin size estimation by 97 subjects reveals a general tendency to underestimate sizes. The specific pattern of results suggests that coin size estimation is influenced by the subject’s attitude towards the coin in question, that inflation and actual value of the coin are not the only determinants of that attitude, and that the attitude remains unchanged if the coin is removed from circulation. Ever since Bruner and Goodman’s work (1947), it has been known that the size of coins, like that of other valuable objects, is usually overestimated. A number of experiments over the years have confirmed this view (McCurdy 1956; Tajfell957; Dawson 1975). The exceedingly high rate of inflation in Israel over the past years (monthly double-digit rate) has created a novel phenomenon: the public has lost its trust in the local currency. The Israeli currency is no longer considered as a store of value, and in many circumstances loses even the functions of means of payment and of unit of account: many prices are expressed and advertised in US dollars, and actual payment is often performed with the US currency. Countless jokes circulate, poking fun at the local currency. These circumstances prompted us to investigate whether the classic money size illusion is to be found in Israel. Two recent studies bear upon this issue. Lea (1981) performed a small scale study in which subjects estimated the size of under their pre- and post-decimalization names. Lea’s own conclusion

pss.sagepub.com

pss.sagepub.com

Address for reprints and correspondence