Difficult Maze Research Articles

A hybrid connectionist/LCS for hidden-state problems

This paper describes and evaluates the performance of a learning classifier system (lcs) inspired algorithm called Temporal Reinforcement And Classification Architecture (traca) on maze navigation tasks which contain hidden state. The evaluation of traca includes comparisons with other learning algorithms on selected difficult maze navigation tasks. Not all lcss are capable of learning all types of hidden-state mazes so traca is specifically compared against selected other lcs-based approaches that are most capable on these tasks, including xcsmh, AgentP (G), and AgentP (SA). Each algorithm is evaluated using a maze navigation task that has been identified as among the most difficult due to recurring aliased regions. The comparisons between algorithms include training time, test performance, and the size of the learned rule sets. The results indicate that each algorithm has its own advantages and drawbacks. For example, on the most difficult maze traca’s average steps to the goal are 10.1 while AgentP (G) are 7.87; however, traca requires an average of only 354 training trials compared with 537 for AgentP (G). Following the maze tasks, traca is also tested on two variations in a truck driving task where it must learn to navigate four lanes of slower vehicles while avoiding collisions. The results show that traca can achieve a low number of collisions with relatively few trials (as low as 24 collisions over 5000 time steps after 10,000 training time steps) but may require multiple network construction attempts to achieve high performance.

Individual subject evaluated difficulty of adjustable mazes generated using quantum annealing

In this study, the maze generation using quantum annealing is proposed. We reformulate a standard algorithm to generate a maze into a specific form of a quadratic unconstrained binary optimization problem suitable for the input of the quantum annealer. To generate more difficulty mazes, we introduce an additional cost function Qupdate to increase the difficulty. The difficulty of the mazes was evaluated by the time to solve the maze of 12 human subjects. To check the efficiency of our scheme to create the maze, we investigated the time-to-solution of a quantum processing unit, classical computer, and hybrid solver. The results show that Qupdate generates difficult mazes tailored to the individual. Furthermore, it show that the quantum processing unit is more efficient at generating mazes than other solvers. Finally, we also present applications how our results could be used in the future.

Investigating The Perseverance and Bias Blind Spot Relationship

Although cognitive bias is easily detectable in the judgments and decision making of other people, we often fail to see its influence in our own thoughts and behavior. This “bias blind spot” is present in most individuals, but its magnitude differs from person to person. To better understand individual differences in susceptibility, we examined the relationship between bias blind spot and perseverance. To measure perseverance, participants were asked to complete as many easy or difficult mazes as they were able to do. Bias blind spot was measured using a validated scale developed by Scopelliti et al. (2015). Bias blind spot susceptibility was compared to time spent on the mazes and number of mazes completed. ANCOVA results revealed a significant finding between bias blind spot and measures of perseverance on difficult mazes, suggesting a relationship exists for bias blind spot and perseverance on challenging tasks. A second study will measure differences in perseverant behavior when feedback is introduced that compares the participant’s number of mazes completed to the average number completed by peers.

Designing Gaze-Based Interactions for Teleoperation: Eye Stick and Eye Click

Perspective-taking and attentional switching are some of the ergonomic challenges that existing teleoperation human-machine interface designs need to address. This study developed two gaze interaction methods, the Eye Stick and the Eye Click, which were based on the joystick metaphor and the navigation metaphor, respectively, to be used in exocentric perspective teleoperation scenarios. We conducted two user studies to test the task performance and the subjective experience of the gaze interaction methods in a virtual ground vehicle teleoperation task. The results showed that compared with a traditional joystick design, the Eye Stick led to a shorter driving distance and the Eye Click led to less task time, and the gaze interaction methods had performance advantages in more difficult mazes. After multiple task sessions, the participants reported that the gaze interaction methods and the traditional joystick were similar in terms of task workload, perceived learnability, and satisfaction; however, the perceived usability of the Eye Stick was not as good as the Eye Click and the traditional joystick. In conclusion, both the Eye Stick and the Eye Click are feasible and promising gaze interaction methods for teleoperation applications with task performance advantages; however, more research is needed to optimize their user experience design.

Aesthetic Science of Rhinoplasty: Three Principles

Aesthetic science is a rapidly growing discipline that involves plastic surgery as much or more than other disciplines of science. Fortunately, how and why the human brain responds to the visual world is better understood now than ever before. Of necessity plastic surgery is a part of that aesthetic science movement because it deals directly with the application of aesthetic principles. Therefore, it behooves the surgeon to be familiar with the principles of aesthetic science so as to both delineate and attain the aesthetic goal at the same time any functional goal is part of the operation. There are three important principles for rhinoplasty: (1) symmetry, (2) continuity and (3) persistence. This study provides examples of how they apply in the preop, intraop and postop phase of performing aesthetic rhinoplasty. A firm understanding of these principles can then guide the surgeon through the difficult maze of decision making, operative planning and surgical execution that arguably makes aesthetic rhinoplasty the most difficult aesthetic operation in plastic surgery. LEVEL OF EVIDENCE V: This journal requires that authors assign a level of evidence to each article. For a full description of these Evidence-Based Medicine ratings, please refer to the Table of Contents or the online Instructions to Authors www.springer.com/00266 .

New books & media

New books & media

The Clinician as Leader: Why, How, and When

Clinicians are inveterate leaders. We lead patients through the difficult maze of illness, families through the travails of ill loved ones, and physicians-in-training through the gauntlet of learning medicine. Yet, in the context of a range of leadership styles that effective leaders must be able to deploy situationally, physician leaders have traditionally defaulted to a "command and control" style that fosters the concept of physicians as "Viking warriors" or "heroic lone healers." The perverse effects of "command and control" are that this style conspires against collaboration and tends to be perpetuated as aspiring leaders emulate their predecessors. Because healthcare faces challenges of cost, access, and quality and is in the throes of change, the current landscape requires effective leadership. Though still relatively uncommon among healthcare organizations, frontrunner organizations are offering leadership development programs. The design of such programs requires clarity about requisite leadership competencies and about how and when to best to deliver such curricula. As one example, the American Thoracic Society has launched its Emerging Leaders Program (ELP), which is currently offering a leadership development curriculum to 18 selected emerging leaders. The ATS ELP curriculum focuses on awareness of self and system and incorporates highly participatory sessions on emotional intelligence, teambuilding, change management, situational leadership, appreciative inquiry, process and quality improvement, strategic planning, and organizational culture. Short-term deliverables are the development and presentation of business plans for innovations proposed by the group. Hoped for longer-term outcomes include an enhanced leadership pipeline for global respiratory health.

Orangutans (Pongo pygmaeus pygmaeus) and children (Homo sapiens) use stick tools in a puzzle box task involving semantic prospection

This study compared three captive orangutans and a group of 5-10 year-old children in their ability to use stick tools to solve a series of mazes in a puzzle box, including three puzzles that required semantic prospection. The puzzle box had seven levels and moveable plastic inserts that created three easy, three intermediate, and three difficult maze configurations. Three wood and three plastic stick tools were presented with each maze. All 26 children immediately solved the easy and intermediate mazes. Seventy-nine percent of the children solved the difficult mazes on their first attempt, and nearly all the children solved the difficult mazes on the second attempt, which suggested a majority of children engaged in effective planning. Girls took significantly longer to solve the intermediate mazes while boys took significantly longer to solve the difficult mazes. Two of three orangutans also successfully avoided the dead ends in the difficult mazes and consistently used stick tools to move peanuts to the goal slots, and took longer to solve the intermediate or difficult mazes. Both the children and orangutans preferred to use plastic tools, although both tool types were functional. These results suggest many similarities between orangutans and children’s abilities to use tools in a puzzle box task that requires planning to avoid dead ends.

Integrative Oncology: An Overview

Integrative oncology, the diagnosis-specific field of integrative medicine, addresses symptom control with nonpharmacologic therapies. Known commonly as "complementary therapies" these are evidence-based adjuncts to mainstream care that effectively control physical and emotional symptoms, enhance physical and emotional strength, and provide patients with skills enabling them to help themselves throughout and following mainstream cancer treatment. Integrative or complementary therapies are rational and noninvasive. They have been subjected to study to determine their value, to document the problems they ameliorate, and to define the circumstances under which such therapies are beneficial. Conversely, "alternative" therapies typically are promoted literally as such; as actual antitumor treatments. They lack biologic plausibility and scientific evidence of safety and efficacy. Many are outright fraudulent. Conflating these two very different categories by use of the convenient acronym "CAM," for "complementary and alternative therapies," confuses the issue and does a substantial disservice to patients and medical professionals. Complementary and integrative modalities have demonstrated safety value and benefits. If the same were true for "alternatives," they would not be "alternatives." Rather, they would become part of mainstream cancer care. This manuscript explores the medical and sociocultural context of interest in integrative oncology as well as in "alternative" therapies, reviews commonly-asked patient questions, summarizes research results in both categories, and offers recommendations to help guide patients and family members through what is often a difficult maze. Combining complementary therapies with mainstream oncology care to address patients' physical, psychologic and spiritual needs constitutes the practice of integrative oncology. By recommending nonpharmacologic modalities that reduce symptom burden and improve quality of life, physicians also enable patients to play a role in their care. Critical for most patients, this also improves the physician-patient relationship, the quality of cancer care, and the well-being of patients and their families.

Spatial memory and executive functions in children

A computerized hidden pathway maze-learning task was used to examine the development of spatial memory and executive functions in 6- to 9-year-olds. Pathway length was manipulated to investigate the impact of increases in maze matrix size on these abilities. Analysis showed that maze matrix size (and ipso facto pathway length) and age interacted to affect executive functions but not spatial memory. Executive errors differed as a function of age on the most difficult maze. Results are discussed in terms of factors affecting the development of executive functions and spatial memory. Three of the authors of this manuscript are employees of CogState Ltd., who own and distribute the GMLT software.

Aging and electrocortical response to error feedback during a spatial learning task

Event-related potentials were collected as older and younger adults responded to error feedback in an adaptation of the Groton Maze Learning Test, an age-sensitive measure of spatial learning and executive skills expected to maximally involve anterior cingulate cortex (ACC). Older adults made more errors and produced smaller feedback-related negativities (FRNs) than young controls. LORETA source localization revealed that, for young adults, neural activation associated with the FRN was focused in ACC and was stronger to negative feedback. Older adults responded with less intense and less differentiated ACC activation, but FRN amplitudes did relate to error rate for the most difficult mazes. The feedback P3 was sensitive to negative feedback but played no role in the prediction of error for either group. These data reflect the selective age-related decline of ACC response but also its continued contribution to performance monitoring in aging.

Carpe Diem: Getting Started on Newspaper Digitization

Historic newspaper digitization is rapidly expanding, yet for the digitization novice, the prospect of launching such a project can be daunting. It is, to say the least, a large and complex initiative to undertake. This paper creates a road map through this difficult maze. It describes the broader landscape, outlines organizational items, and discusses the pros and cons of selecting content and source materials, metadata, article segmentation, copyright, and other technical issues.

Gradient Descent Methods in Learning Classifier Systems: Improving XCS Performance in Multistep Problems

The accuracy-based XCS classifier system has been shown to solve typical data mining problems in a machine-learning competitive way. However, successful applications in multistep problems, modeled by a Markov decision process, were restricted to very small problems. Until now, the temporal difference learning technique in XCS was based on deterministic updates. However, since a prediction is actually generated by a set of rules in XCS and Learning Classifier Systems in general, gradient-based update methods are applicable. The extension of XCS to gradient-based update methods results in a classifier system that is more robust and more parameter independent, solving large and difficult maze problems reliably. Additionally, the extension to gradient methods highlights the relation of XCS to other function approximation methods in reinforcement learning.

Personal Accounts: From the Depths of Despair to a Mission of Advocacy

Personal Accounts: From the Depths of Despair to a Mission of Advocacy

Effects of puzzle feeders on pathological behavior in individually housed rhesus monkeys.

Self-injurious behavior (SIB) occurs in about 10% of individually housed monkeys. Monkeys with SIB bite their own bodies frequently, occasionally inflicting wounds as a result. At present, there is no standard treatment for this phenomenon. We examined the effectiveness of puzzle feeders in alleviating SIB in monkeys with a veterinary record of self-inflicted wounding. Two groups of monkeys (SIB and controls) were exposed to puzzle feeders for a 6 week period. Three levels of maze difficulty were examined. All monkeys used the feeders, but manipulation was confined to a brief period immediately after the feeders were loaded each day (1000 h) and was infrequent during the later sampling periods (1100 and 1400 h). The most difficult maze yielded a slight increase in usage at 1100 h. During the puzzle feeder phase, whole body stereotypies, including pacing and rocking, were reduced substantially in all monkeys at 1000 h when feeder manipulation was at its highest. However, self-biting in the SIB group was unchanged. Some monkeys actually bit themselves while manipulating the feeder. Long-term effects on abnormal behavior were assessed by comparing behavior during the feeder phase to baseline periods and to a phase in which the monkeys were provisioned with treats placed directly into their food box. Whole body stereotypies, including pacing, were reduced during both treatment phases; however, the reduction was associated only with the 1000 h observation. Puzzle feeders were more effective than treats alone in alleviating whole body stereotypies. Self-biting was unchanged through all phases. Puzzle feeders are beneficial from the perspective of eliciting manipulation. They also yield transient reductions in whole body stereotypy, an effect that does not extend beyond the direct manipulation of the feeder. Puzzle feeders are ineffective in alleviating self-injurious behavior.

Behavioral impairment in radial-arm maze learning and acetylcholine content of the hippocampus and cerebral cortex in aged mice

Age-related changes in spatial learning performance were studied in relation to acetylcholine (ACh) content of brain regions in male aged (28-month-old) and young (5-month-old) mice of BDF1 strain. As there were large individual differences in the spatial performance of aged mice, the aged mice were divided into two subgroups, old A and old B. The old A group included the six best performers out of the 12 aged mice and the old B group included the remaining 6 worst performers. In a radial-arm maze task with 8 baited arms, aged mice in the old B group showed a marked deficit in acquisition performance and habituation to the apparatus. In the more difficult maze task with only 4 baited arms, the aged mice in the old B group exhibited marked impairment both in working memory and reference memory throughout training, whereas the aged mice in the old A group showed deficits in reference memory during the first 20 days of training and working memory during the last 20 days relative to young mice. Neurochemical analysis revealed significant decreases in the ACh content of the hippocampus and striatum in both aged groups, and in the frontal cortex and posterior cortex of the old B group as compared to the young group. Correlational analysis showed significant correlations between learning performance in the spatial task and ACh levels in the hippocampus, frontal cortex, and posterior cortex. The present results demonstrate that a marked decrease in ACh levels in the hippocampus and cerebral cortex may be closely related to behavioral impairment in spatial learning and habituation in fully aged mice. Thus, this particular strain of mice may serve as a useful animal model for age-related impairment of cognitive functions in relation to hypofunction of the cholinergic systems in the basal forebrain.

Developmental and Task Factors in Logo Programming

Ninety children (7 to 12 years) were taught simplified Logo programming commands, which they used to solve computer mazes. Mazes were solved under conditions of either immediate—each command entered had an immediate, visible effect on the movement of the turtle—or delayed-feedback—the invisible turtle moved with each command, but showed its current position only after every six commands. These conditions were designed to model programming interactively versus in the editor. All children solved a final maze with no-feedback (i.e., the turtle moved invisibly, but never showed its position). It was hypothesized that experience in immediate mode would facilitate learning in younger children, but that experience in delayed mode would lead to equal or better learning in older children. Performance was scored for programming errors. Older children performed better than younger children on all mazes; performance with immediate feedback was better than that with delayed; and more difficult mazes tended to elicit higher error rates or longer latencies. On the final no-feedback maze, prior experience in delayed mode led to poor performance for seven-eight-year olds, but not for older children. Near transfer of programming performance was examined with two off-computer tasks (i.e., command generation and command interpretation). In correlational analyses measures of mental capacity (i.e., working memory) and analytical-ability/field-independence were found to relate to programming performance. Findings have implications for the timing of introduction of the editor in the Logo curriculum.

The effects of an intramaze cue search rule on rat's spatial working memory

Two groups of rats were trained in an elevated 8-arm radial maze consisting of four smooth and four hardware cloth covered (grid) arms. Rats in one group (Texture Relevant) had to enter four baited arms of one texture, Set A, before they could be rewarded for entering the second set of arms of the other texture, Set B. Rats in the other group (Texture Irrelevant) were not constrained in their pattern of arm selection. Rats in both groups were run in the maze until they had sampled all baited arms. Although Texture Relevant (TR) rats initially made more reentries than Texture Irrelevant (TI) rats during the first training phase, rats in both groups reduced reentries to equivalent, low asymptotic levels. TR rats also made many premature entries to Set B arms during initial training but reduced these errors with continued training. When adjacent arms became less spatially differentiated with reference to extramaze room stimuli in the second phase and in a replication of this experiment, TI rats increased reentries to earlier sampled arms more than TR rats. TR rats, however, increased premature entries into Set B when arms within each texture set became less spatially differentiated with respect to extramaze room stimuli. These results indicate that TR rats developed a stable hierarchical representation of the maze, which enhanced their retention of earlier visited arms in more difficult maze configurations.

The Hebb-Williams test to assess recovery of learning after limbic lesions in mice

A scaled-down Hebb-Williams closed-field maze was constructed and adapted for use with mice. In a first experiment, the acquisition performance of the original 12 Hebb-Williams mazes by BALB/c mice was studied in order to re-group the mazes into 3 main groups of difficulty (low, moderate and high). In a second experiment different problems from each class of difficulty were presented at 3 times intervals (between 3 and 7 weeks after surgery) to mice with lesions of the cingulate cortex. The rate of acquisition in lesioned mice was observed to be facilitated between 3 and 5 weeks after surgery for the most difficult mazes. This effect was reversed (impairment) at 7 weeks post-lesion. The observed biphasic behavioral modifications constitute evidence that our present protocol provides a useful tool to evaluate the time-course of changes in learning performance following brain injury.

Effects of protein-calorie deficiencies on the learning ability of the wistar rat

The present experiment employed the deprivation model. The treatment conditions were defined by the severity of protein-calorie deficiencies and the growth-developmental period at which such deficiencies were imposed. Prenatal and neonatal deficiencies were imposed by maintaining the mothers on the diets during pregnancy or lactation. Early deficiency was accomplished by maintaining weaned pups on their respective diets from 21–42 days of age. After realimentation on an adequate diet, the experimental animals were tested in three water mazes. Protein-calorie deficiencies did not cause permanent reduction in body weight, but neonatal deficiencies were found to impair the learning of the most difficult maze. These and other findings were discussed in the context of physiological changes which might accompany protein deficiencies, and it was concluded that the permanence of the effects on learning ability was related to the period of growth and development at which the deficiencies occur, the severity of the deficiencies, and the difficulty of the learning task.