Age-Based Differences in Strategy Use in Choice Tasks

Abstract

We incorporated behavioral and computational modeling techniques to examine age-based differences in strategy use in two four-choice decision-making tasks. Healthy older (aged 60–82 years) and younger adults (aged 18–23 years) performed one of two decision-making tasks that differed in the degree to which rewards for each option depended on the choices made on previous trials. In the choice-independent task rewards for each choice were not affected by the sequence of previous choices that had been made. In contrast, in the choice-dependent task rewards for each option were based on how often each option had been chosen in the past. We compared the fits of a model that assumes the use of a win-stay–lose-shift (WSLS) heuristic to make decisions, to the fits of a reinforcement-learning (RL) model that compared expected reward values for each option to make decisions. Younger adults were best fit by the RL model, while older adults showed significantly more evidence of being best fit by the WSLS heuristic model. This led older adults to perform worse than younger adults in the choice-independent task, but better in the choice-dependent task. These results coincide with previous work in our labs that also found better performance for older adults in choice-dependent tasks (Worthy et al., 2011), and the present results suggest that qualitative age-based differences in the strategies used in choice tasks may underlie older adults’ advantage in choice-dependent tasks. We discuss possible factors behind these differences such as neurobiological changes associated with aging, and increased use of heuristics by older adults.

Altered behavioral and neural responsiveness to counterfactual gains in the elderly

“…In addition, the elderly adults demonstrated a similar FPE+ response in the right ventral striatum as the young, and the FPEQ model weighted FPE+ similarly for both groups, showing no differences in behavioral or neural responsiveness to counterfactual losses between groups. Previous research has demonstrated that healthy elderly adults are less affected by counterfactual outcomes (Brassen et al, 2012) and actually utilize decision making strategies that don't rely on reward predictions, such as win-stay-loseshift (Worthy & Maddox, 2012). The results of the model fitting shows that elderly adults SSIT performance was more likely to be guided by reward predictions than a strategy that doesn't utilize reward predictions.…”

“…Indeed, vmPFC activation is correlated with expected values, as it is in young adults, for older adults who perform more optimally (SamanezLarkin et al, 2011). Moreover, advanced age is associated with different decision-making strategies depending on the task contingencies (Worthy & Maddox, 2012) and older adults choice behavior is sometimes better explained by a win-staylose-shift (WSLS) strategy that does not rely on a reward prediction (Worthy & Maddox, 2012), as opposed to reinforcement learning strategies that do. This supports the notion that although the elderly use different decision-making strategies dependent on the task context, the vmPFC is important for processing reward predictions to guide choice behavior in some contexts.…”

Decision Making Across the Life Span

“…An example of this is highlighted in the paper by Worthy and Maddox (2012) that uses computational models of learning and decision making to identify strategy differences between younger and older adults. Consistent with previous studies of age differences in strategy use , Worthy and Maddox (2012) show that older adults use a simpler strategy (win-stay/lose-shift) compared to younger adults (who are better fit by a more traditional reinforcement learning model).…”

“…The papers also cover a broad range of topics including reward effects on learning and memory, risky decision making, intertemporal choice, strategy use, and financial decision making in healthy adults (Gilbert et al, 2011;Mather and Schoeke, 2011;Samanez-Larkin et al, 2011a;Cavanagh et al, 2012;Mata et al, 2012;Shivapour et al, 2012;Spaniol and Wegier, 2012;Westbrook et al, 2012;Worthy and Maddox, 2012) along with a complementary study on susceptibility to misleading advertisements in individuals with frontal cortical damage .…”

“…Here we feature empirical work ranging from behavioral (Mather and Schoeke, 2011;Cavanagh et al, 2012;Shivapour et al, 2012;Spaniol and Wegier, 2012;Westbrook et al, 2012;Worthy and Maddox, 2012) and computational (Samanez-Larkin et al, 2011a;Cavanagh et al, 2012;Worthy and Maddox, 2012) to cognitive neuroscience (Samanez-Larkin et al, 2011a) and non-human animal research (Gilbert et al, 2011), investigating age similarities and differences in decision making, together with theoretical perspectives (Mata et al, 2012) that integrate existing evidence and provides new insights. The papers also cover a broad range of topics including reward effects on learning and memory, risky decision making, intertemporal choice, strategy use, and financial decision making in healthy adults (Gilbert et al, 2011;Mather and Schoeke, 2011;Samanez-Larkin et al, 2011a;Cavanagh et al, 2012;Mata et al, 2012;Shivapour et al, 2012;Spaniol and Wegier, 2012;Westbrook et al, 2012;Worthy and Maddox, 2012) along with a complementary study on susceptibility to misleading advertisements in individuals with frontal cortical damage .…”