description9 papers

group7 followers

lightbulbAbout this topic

We investigate the hypothesis that historical information plays an important role in learning action selection via reinforcement learning. In particular, we consider the value of the history of prior actions in the classic T maze of...more

We investigate the hypothesis that historical information plays an important role in learning action selection via reinforcement learning. In particular, we consider the value of the history of prior actions in the classic T maze of Tolman and Honzik (Tolman & Honzik 1930). We show that including a sequence of actions in the state makes it possible to learn the task using reinforcement learning. Moreover we show that learning over sequences of length 0 ~ 4 is necessary to model rat behavior. This behavior is modeled in Soar-RL and compared to an earlier model created in ACT-R.

Reinforcement learning (RL) models of decision-making cannot account for human decisions in the absence of prior reward or punishment. We propose a mechanism for choosing among available options based on goal-option association strengths,...more

Reinforcement learning (RL) models of decision-making cannot account for human decisions in the absence of prior reward or punishment. We propose a mechanism for choosing among available options based on goal-option association strengths, where association strengths between objects represent object proximity. The proposed mechanism, Goal-Proximity Decision-making (GPD), is implemented within the ACT-R cognitive framework. A one-choice navigation experiment is presented. GPD captures human performance in the early trials of the experiment, where RL cannot.

There are many varieties of task analysis-each with its advantages and disadvantages, each with its adherents and detractors (e.g. see the recent collections published by Annett & Stanton, 1998; Kirwan & Ainsworth, 1992). Most published...more

There are many varieties of task analysis-each with its advantages and disadvantages, each with its adherents and detractors (e.g. see the recent collections published by Annett & Stanton, 1998; Kirwan & Ainsworth, 1992). Most published descriptions focus on how to apply the technique or why it is a good technique to apply. Few accounts written by advocates of a technique are specifically directed at problems and pitfalls in applying the technique. This account is different. Although we are unabashedly enthusiastic advocates of the theory-driven combination of task analysis and protocol analysis that we employ, we hope that by identifying problems and obstacles that we encountered that more people will be better prepared and, therefore, more successful at applying these techniques. Beware-knowing that the road is narrow, winding, and unmarked does not make the trip easy. It might, however, discourage someone from setting out in the family sedan. For those who are better equipped, knowledge of the hazards ahead may help them avoid blindly plunging forward into a known problem. It is in this spirit that we write this chapter. The following section provides a brief overview of the techniques we employ. It then introduces the known obstacles to these techniques. The main part of the chapter discusses these obstacles in the context of a specific project-Project Nemo. Theory-Driven Task Analysis and Protocol Analysis Theory-driven task analysis decomposes the procedural and declarative knowledge required to perform a task into components supported by the theory. With some additional work on the part of the analyst, the control structure provided by the theory can use the elements of the analysis Gray & Kirschenbaum page 3 Analyzing a Novel Expertise to form a model of how a user performs the task. Theories with weak or rigid control structures, such as keystroke-level GOMS or CPM-GOMS (for an overview, see John & Kieras, 1996a; John & Kieras, 1996b), may produce models that are only capable of performing the exact task that was analyzed. Theories with more powerful control structures, such as NGOMSL, ACT-R, Soar, or EPIC (see Gray, Young, & Kirschenbaum, 1997b), may respond adaptively to perform variations of the analyzed task. Cognitive theories provide constraints to the final form of the analysis-that is, for how the components must fit together. However, the components per se vary widely, and the analysis of

On the basis of findings from an experiment with 6-year-old children we show a proposal for a cognitive model of representational shifts in learning the number line. The findings from the experiment provide information on number line...more

On the basis of findings from an experiment with 6-year-old children we show a proposal for a cognitive model of representational shifts in learning the number line. The findings from the experiment provide information on number line estimation-that is, translating a number to a spatial position on a number line. Though the experiment is a replication of an experiment done by Siegler and Ramani (2008) where they concluded with a logarithmic to linear shift, we could not find logarithmic representation of the results from any of our subjects. What we find is anchor points as important for improvement on learning the number line.

We investigate the hypothesis that historical information plays an important role in learning action selection via reinforcement learning. In particular, we consider the value of the history of prior actions in the classic T maze of...more

We investigate the hypothesis that historical information plays an important role in learning action selection via reinforcement learning. In particular, we consider the value of the history of prior actions in the classic T maze of Tolman and Honzik (Tolman & Honzik 1930). We show that including a sequence of actions in the state makes it possible to learn the task using reinforcement learning. Moreover we show that learning over sequences of length 0 ~ 4 is necessary to model rat behavior. This behavior is modeled in Soar-RL and compared to an earlier model created in ACT-R.

Reinforcement learning (RL) models of decision-making cannot account for human decisions in the absence of prior reward or punishment. We propose a mechanism for choosing among available options based on goal-option association strengths,...more

Reinforcement learning (RL) models of decision-making cannot account for human decisions in the absence of prior reward or punishment. We propose a mechanism for choosing among available options based on goal-option association strengths, where association strengths between objects represent object proximity. The proposed mechanism, Goal-Proximity Decision-making (GPD), is implemented within the ACT-R cognitive framework. A one-choice navigation experiment is presented. GPD captures human performance in the early trials of the experiment, where RL cannot.

Cognitive engineering is the application of cognitive science theories to human factors practice. As this description suggests, there are strong symbioses between cognitive engineering and cognitive science, but there are also strong...more

Cognitive engineering is the application of cognitive science theories to human factors practice. As this description suggests, there are strong symbioses between cognitive engineering and cognitive science, but there are also strong differences.

Symbiosis implies a mutual influence, and the history of cognitive engineering supports this characterization in two key areas: the development of cognitive theory and the development of computational modeling software. For theory development, a stringent test of our understanding of cognitive processes is whether we can apply our knowledge to real-world problems. The degree to which we succeed at this task is the degree to which we have developed robust and powerful theories. The degree to which we fail at this task is the degree to which more research and stronger theories are required (Gray, Schoelles, & Myers, 2004).

A theoretical analysis of the development of numerical representations indicated that playing linear number board games should enhance preschoolers' numerical knowledge and ability to acquire new numerical knowledge. The effect on...more

A theoretical analysis of the development of numerical representations indicated that playing linear number board games should enhance preschoolers' numerical knowledge and ability to acquire new numerical knowledge. The effect on knowledge of numerical magnitudes was predicted to be larger when the game was played with a linear board than with a circular one, due to a more direct mapping between the linear board and the desired mental representation. As predicted, playing the linear board game for roughly one hour increased lowincome preschoolers' proficiency on the two tasks that directly measured understanding of numerical magnitudes -numerical magnitude comparison and number line estimation -more than playing the game with a circular board or engaging in other numerical activities. Also as predicted, children who had played the linear number board game generated more correct answers and better quality errors in response to subsequent training on arithmetic problems, a task hypothesized to be influenced by knowledge of numerical magnitudes. Thus, playing linear number board games not only increases preschoolers' numerical knowledge; it also helps them learn from future numerical experiences.

Theoretical analyses of the development of numerical representations suggest that playing linear number board games should enhance young children's numerical knowledge. Consistent with this prediction, playing such a game for roughly 1 hr...more

Theoretical analyses of the development of numerical representations suggest that playing linear number board games should enhance young children's numerical knowledge. Consistent with this prediction, playing such a game for roughly 1 hr increased low-income preschoolers' (mean age 5 5.4 years) proficiency on 4 diverse numerical tasks: numerical magnitude comparison, number line estimation, counting, and numeral identification. The gains remained 9 weeks later. Classmates who played an identical game, except for the squares varying in color rather than number, did not improve on any measure. Also as predicted, home experience playing number board games correlated positively with numerical knowledge. Thus, playing number board games with children from low-income backgrounds may increase their numerical knowledge at the outset of school.

Children with mild intellectual disabilities (MID) appear to have particular problems in understanding the numerical meaning of Arabic digits. Therefore, we developed and evaluated a numerical domino game that specifically targeted the...more

Children with mild intellectual disabilities (MID) appear to have particular problems in understanding the numerical meaning of Arabic digits. Therefore, we developed and evaluated a numerical domino game that specifically targeted the association between these digits and the numerical magnitudes they represent. Participants were 30 children with MID (M = 8.36 years), randomly assigned to either the numerical domino game or to a control color domino game. Findings revealed that both groups of children improved on a nonsymbolic comparison and arithmetic task. Most importantly, only children who played the numerical domino game became significantly faster from pretest to posttest on a symbolic comparison task. These findings suggest that numerical magnitude processing can be successfully trained in children with MID.

Planning a path to a destination, given a number of options and obstacles, is a common task. We suggest a two-component cognitive model that combines retrieval of knowledge about the environment with search guided by visual perception. In...more

Planning a path to a destination, given a number of options and obstacles, is a common task. We suggest a two-component cognitive model that combines retrieval of knowledge about the environment with search guided by visual perception. In the first component, subsymbolic information, acquired during navigation, aids in the retrieval of declarative information representing possible paths to take. In the second component, visual information directs the search, which in turn creates knowledge for the first component. The model is implemented using the ACT-R cognitive architecture and makes realistic assumptions about memory access and shifts in visual attention. We present simulation results for memory-based high-level navigation in grid and tree structures, and visual navigation in mazes, varying relevant cognitive (retrieval noise and visual finsts) and environmental (maze and path size) parameters. The visual component is evaluated with data from a multi-robot control experiment, where subjects planned paths for robots to explore a building. We describe a method to compare trajectories without referring to aligned points in the itinerary. The evaluation shows that the model provides a good fit, but also that planning strategies may vary with task loads.

A theoretical analysis of the development of numerical representations indicated that playing linear number board games should enhance preschoolers' numerical knowledge and ability to acquire new numerical knowledge. The effect on...more

A theoretical analysis of the development of numerical representations indicated that playing linear number board games should enhance preschoolers' numerical knowledge and ability to acquire new numerical knowledge. The effect on knowledge of numerical magnitudes was predicted to be larger when the game was played with a linear board than with a circular one, due to a more direct mapping between the linear board and the desired mental representation. As predicted, playing the linear board game for roughly one hour increased lowincome preschoolers' proficiency on the two tasks that directly measured understanding of numerical magnitudes -numerical magnitude comparison and number line estimation -more than playing the game with a circular board or engaging in other numerical activities. Also as predicted, children who had played the linear number board game generated more correct answers and better quality errors in response to subsequent training on arithmetic problems, a task hypothesized to be influenced by knowledge of numerical magnitudes. Thus, playing linear number board games not only increases preschoolers' numerical knowledge; it also helps them learn from future numerical experiences.

On the basis of findings from an experiment with 6-year-old children we show a proposal for a cognitive model of representational shifts in learning the number line. The findings from the experiment provide information on number line...more

On the basis of findings from an experiment with 6-year-old children we show a proposal for a cognitive model of representational shifts in learning the number line. The findings from the experiment provide information on number line estimation -that is, translating a number to a spatial position on a number line. Though the experiment is a replication of an experiment done by where they concluded with a logarithmic to linear shift, we could not find logarithmic representation of the results from any of our subjects. What we find is anchor points as important for improvement on learning the number line.

We present findings from an experiment with 6-year-old children in Norway who are getting started with the process of learning the numbers from 1 to 10. The findings provide information on number line estimation -that is, translating a...more

We present findings from an experiment with 6-year-old children in Norway who are getting started with the process of learning the numbers from 1 to 10. The findings provide information on number line estimation -that is, translating a number to a spatial position on a number line. The results show different categories of representation of the magnitudes on the number line, which may represent different stages in a learning sequence. On this basis, we show a proposal for a cognitive model of the learning process towards a linear representation of magnitudes.

Semantic Relevancy Maps are a visual analytic technique for representing the distribution of semantic relevancy across an information display. The maps highlight the text areas of the display corresponding to the relevance of that text to...more

Semantic Relevancy Maps are a visual analytic technique for representing the distribution of semantic relevancy across an information display. The maps highlight the text areas of the display corresponding to the relevance of that text to user goals, with stronger highlights indicating higher degrees of relevance. Semantic Relevancy Maps were developed as a tool for high-fidelity computational cognitive models that search complex information displays in the same manner as humans. However, they offer the potential to be a standalone tool for quickly evaluating the spatial layout of information for designers or, more simply, for identifying the spatial location of sought-for information by any computer user.

Abstract When people search a Web page for links that are relevant to their information goal, they attend to the labels and estimate the likelihood that the link will lead to the goal. We have previously found in a simplified single-page...more

Abstract When people search a Web page for links that are relevant to their information goal, they attend to the labels and estimate the likelihood that the link will lead to the goal. We have previously found in a simplified single-page menu search task that people sometimes, but not always, assess only a subset of the links available.

Abstract Planning a path to a destination, given a number of options and obstacles, is a common task. We suggest a two-component cognitive model that combines retrieval of knowledge about the environment with search guided by visual...more

Abstract Planning a path to a destination, given a number of options and obstacles, is a common task. We suggest a two-component cognitive model that combines retrieval of knowledge about the environment with search guided by visual perception. In the first component, subsymbolic information, acquired during navigation, aids in the retrieval of declarative information representing possible paths to take.

Successfully explaining and replicating the complexity and generality of human and animal learning will require the integration of a variety of learning mechanisms. Here we introduce a computational model which integrates associative...more

Successfully explaining and replicating the complexity and generality of human and animal learning will require the integration of a variety of learning mechanisms.  Here we introduce a computational model which integrates associative learning and reinforcement learning. We contrast the integrated model with standalone associative learning and reinforcement learning models in three simulation studies. First, a synthetic grid-navigation task is employed to highlight performance advantages for the integrated model in an environment where the reward structure is both diverse and dynamic. The second and third simulations contrast the performances of the three models in behavioral experiments, demonstrating advantages for the integrated model in accounting for behavioral data.

Reinforcement learning (RL) models of decision-making cannot account for human decisions in the absence of prior reward or punishment. We propose a mechanism for choosing among available options based on goal-option association strengths,...more

Reinforcement learning (RL) models of decision-making cannot account for human decisions in the absence of prior reward or punishment. We propose a mechanism for choosing among available options based on goal-option association strengths, where association strengths between objects represent previously experienced object proximity. The proposed mechanism, Goal-Proximity Decision-making (GPD), is implemented within the ACT-R cognitive framework. GPD is found to be more efficient than RL in three maze-navigation simulations. GPD advantages over RL seem to grow as task difficulty is increased. An experiment is presented where participants are asked to make choices in the absence of prior reward. GPD captures human performance in this experiment better than RL.

We applied overlapping waves theory and microgenetic methods to examine how children improve their estimation proWciency, and in particular how they shift from reliance on immature to mature representations of numerical magnitude. We also...more

We applied overlapping waves theory and microgenetic methods to examine how children improve their estimation proWciency, and in particular how they shift from reliance on immature to mature representations of numerical magnitude. We also tested the theoretical prediction that feedback on problems on which the discrepancy between two representations is greatest will cause the greatest representational change. Second graders who initially were assessed as relying on an immature representation were presented feedback that varied in degree of discrepancy between the predictions of the mature and immature representations. The most discrepant feedback produced the greatest representational change. The change was strikingly abrupt, often occurring after a single feedback trial, and impressively broad, aVecting estimates over the entire range of numbers from 0 to 1000. The Wndings indicated that cognitive change can occur at the level of an entire representation, rather than always involving a sequence of local repairs.

We applied overlapping waves theory and microgenetic methods to examine how children improve their estimation proWciency, and in particular how they shift from reliance on immature to mature representations of numerical magnitude. We also...more

We applied overlapping waves theory and microgenetic methods to examine how children improve their estimation proWciency, and in particular how they shift from reliance on immature to mature representations of numerical magnitude. We also tested the theoretical prediction that feedback on problems on which the discrepancy between two representations is greatest will cause the greatest representational change. Second graders who initially were assessed as relying on an immature representation were presented feedback that varied in degree of discrepancy between the predictions of the mature and immature representations. The most discrepant feedback produced the greatest representational change. The change was strikingly abrupt, often occurring after a single feedback trial, and impressively broad, affecting estimates over the entire range of numbers from 0 to 1000. The Wndings indicated that cognitive change can occur at the level of an entire representation, rather than always involving a sequence of local repairs.