Eric J. Johnson - US grants

EIA-0140304 -Salvatore Stolfo-Columbia University-SGER: Mitigating Access Risks of Browsing Government Data and Websites by Secure Private Portals

This proposal from a computer scientist and a social scientist will study the social and policy implications and citizen perceptions of government gathering of Personally Identifying Information through Federal web sites, and will explore the use of alternative privacy-enhancing technologies. These technologies will be evaluated and further developed. Initial discussions with several interested Federal Government agencies have already begun.

The proposal examines preference, a key construct in decision science and economics, as the natural output of the human memory system, following the properties and characteristics of other types of knowledge. This preferences-as-memory approach suggests that preferences are neither "constructed" from first principles anew on each occasion, nor completely stable and immutable. The research attempts to explain and predict important phenomena in which preference deviates from rational-economic model prescription (including loss aversion, intertemporal discounting, and tradeoff difficulty) by examining the memory mechanisms involve in preference construction. These include priming, i.e., a transient increase in the accessibility of a concept after presentation of a related concept; inhibition or interference, i.e., a class of phenomena where instructions to recall parts of previously learned materials hinders subsequent recall of the rest; reactivity, i.e. where access of memory produces changes in its content and structure; structure of memory representations, i.e., the number of concept attributes and the system of their interconnections.

The proposed theoretical and empirical work has the following three purposes: (1) to convince behavioral decision researchers that they are ignoring the (memory) processes involved in the formation and expression of preference at their peril, (2) to demonstrate that current considerations of mostly explicit memory processes are incomplete, and (3) to show that the incorporation of a small number of well established memory properties provides conceptual integration across a large number of preference and choice phenomena that range from loss aversion to tradeoff difficulty, subadditivity, anchoring, and overconfidence. Societally important practical applications also arise from the use of a memory framework for understanding the formation and modification of preferences. A better understanding of developmental changes in memory processes and preference construction could help, for example, to design interventions that would minimize socially harmful consequences of changes in memory performance on judgment and choice in geriatric populations.

DESCRIPTION (provided by applicant): This proposal argues that we can leverage what we know about memory changes over the lifespan to understand how decision-making changes as people get older. The proposed research concentrates on four important, robust and well-studied decision phenomena that are problematic for younger adults and may be stronger in older adults: The endowment effect, asymmetries in intertemporal choice, default effects, and anchoring. Using an approach to decision-making called Preferences-as- Memories (PAM), we develop models of how people make judgments and choices in these situations and describe the role of memory processes in the process of preference construction, particularly the role of interference, a memory characteristic which changes over the life course. We then propose experiments to test these models using both personally administered and web-based experiments that will contrast the pattern of preferences exhibited by older adults with those of younger adults. Using the web allows one to reach a broad population of older and younger adults more quickly. Personally administered experiments allow for greater experimental control and for a fine-tuning of new methodologies and procedures. Our experiments will employ methods for measuring choices and preferences in well-defined decision-making tasks, along with tools for measuring information retrieval during decision-making both implicitly and explicitly. By further measuring performance in a range of additional cognitive domains, we will also be able to test between alternative mediating mechanisms for the preference phenomena that will be under study. The decision-making phenomena that will be examined have implications for the welfare of older adults. If, for example, older people show greater endowment effects (and therefore greater loss aversion), they may be more attached to some possessions and avoid change, even when this may not be in their best interest. Because the proposed research generates process models of how memory representations and processes affect decisions, it provides directions for the development of interventions that can aid and improve decisions - an important contribution in an aging society that tries to shift important decisions (e.g., on pension investments and health care choices) from government agencies to individuals.

DESCRIPTION (provided by applicant): An intriguing discrepancy exists between the performance deficits traditionally observed in the laboratory and the general competencies that older adults exhibit in everyday life. Parallel to the general public's two conflicting views about aging-that age brings wisdom, and that age brings deteriorating skills-research has found mixed evidence showing that older adults' decisions can be both wiser and less wise compared to their younger counterparts. Convergent evidence suggests that emotional regulation, reasoning about interpersonal conflicts, and crystallized intelligence all improve with advanced age. However, older adults also suffer from declines in the cognitive abilities necessary for processing new information, including reasoning, dividing attention, processing speed, and working memory-i.e., what is known as fluid intelligence. Whereas the effects of declines in fluid intelligence on decisions, particularly novel ones, are well understood, the proposed research will examine what components of crystallized intelligence-the part of general intelligence that involves knowledge, experience, and expertise-contribute to wisdom in decision making in older adults. This application has two specific goals. The first is to clarify the relationship between crystallized intelligence and decision making. Previous work has shown that better crystallized intelligence may compensate for worse fluid intelligence in older adults, but we know little about what abilities the tests of general knowledge and vocabulary used to measures crystallized intelligence actually capture and how these domain-general abilities contribute to decision quality across domains. For example, domain-general knowledge may correlate with domain-specific (i.e., semantic) knowledge, or it may proxy for a number of other abilities that improve with age, including domain-specific process expertise (i.e., automatic processing ability gained from practice), domain-general principles (e.g., knowing to seek advice, eliminating dominated alternatives, and identifying redundant information), and emotion-based aspects of decision making (such as reliance on affective sources of information). The second goal is to explore the possible connection between crystallized intelligence and an important, specific kind of decision skill, affective forecasting, or the ability to predict what will make us happy. Affective variable play a major role in decision processes, and age differences in emotional aspects of decision-making would seem to have important implications for choice outcomes. If older adults make decisions that deviate from what is economically rational, they may nonetheless be happier with their decisions because they are better at predicting how they will feel about the outcomes. Individual differences in affective forecasting accuracy may also be influenced by other abilities that change with age, including components of both fluid and crystallized intelligence. We examine decisions that have implications for the welfare of older adults, across a wide range of domains, including retirement and healthcare. Research insights gained will provide directions for the development of interventions that can aid and improve decisions-an important contribution in an aging society when major decisions (e.g., retirement accumulation and spending and health care choices) are increasingly up to individuals. PUBLIC HEALTH RELEVANCE: By understanding what components of individual and lifespan differences in crystallized intelligence lead adults across the lifespan to make better decisions-both economically and emotionally-across a wide range of domains (including retirement and healthcare), the proposed research provides directions for the development of interventions that can aid decisions and improve the welfare of older adults. These psychological and behavioral economic interventions offer an important contribution in an aging society that is increasingly shifting socially and individually important decisions (e.g., on retirement saving and decumulation and on health care choices) from government agencies to individuals. The tenuous evidence for the effectiveness of fluid intelligence training (Owen et al., 2010) makes it important to find alternative avenues to improve decision quality by designing interventions that capitalize on components of crystallized intelligence that influence competencies relevant to decision and choice.