How do our social choices and attention to them change over our lifespan?

54  Download (0)

Full text


How do our social choices and attention to them change over our lifespan?

An experimental approach to the questions of whether age and attention has an effect on social decisions.


Humans are known to be social beings. Across our lifespan we are constantly making social decisions regarding with whom we wish to spend time. As we grow older, we face different challenges, new opportunities and develop, especially in a social sense. It is, therefore, not strange to expect that our social decisions also change as we age. Another aspect that is known to have an effect on decisions in general is attention. The aim of this study is to

investigate what the effect is of age and attention on our social decisions. In an online mouse- tracking experiment participants were asked to make social decisions. They were introduced to an open-ended and time-limited scenario and were either only given information about the people they could spend time with or some additional information. The results regarding the effect of age on social choice showed that in the additional info open-ended scenario older adults went more for the novel choice. Furthermore, the results regarding the effect of attention on social choice showed a significant effect in the person open-ended and

additional info time-limited scenario, guiding their social choice. Age was shown to have no effect on the attention of the participant. The results are therefore not complete as not all scenarios were significant. Therefore, there was not enough evidence to conclude that age or attention effected the social decisions.

Nicole Koobs (12786179)

MSc Businesss Economics: Neuroeconomics Supervisor: Dianna Amasino

August 2021



While writing this thesis and throughout my education I had a lot of support and guidance. I would especially like to mention a couple of individuals that I am especially grateful to.

First of all, I would like to thank my supervisor, dr Dianna Amasino, for supporting me throughout the process by giving constructive feedback, helping me critically think about the topic and experiment and promptly responding to all my emails and questions. This made the process enjoyable, run smoothly and challenged me throughout the process.

Furthermore, I would like to state my appreciation to the master track coordinator, dr Jan Engelmann, for his support throughout the master program.

In addition, I would like to thank my family, Gaynor McGregor, Helena Koobs, Lucienne Koobs, Gina Koobs and Sam Chadwick for their contribution in data collection and support throughout the thesis process and for their support throughout my studies in general, especially after the challenging year we have had.

Lastly, I would like to thank my friends for their support throughout the past two years with my studies, data collection, critically discussing my thesis and general support during this strange and challenging year. The friends I would especially like to acknowledge are Avalon Schuurman, Danielle Ernesten, Emma Dolman, Tara Parmar, Hoda Alaa El Din and Anne Slagman.

Statement of Originality

This document is written by student Nicole Koobs who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in the document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision and completion of the work, not for the contents.


Table of Contents

1. Introduction ...4

2. The ageing brain ...6

2.1 Volume ...6

2.2 Grey matter ...6

2.3 White matter ...7

2.4 Hippocampus ...8

2.5 Prefrontal cortex ...9

3. Social preferences over the life span ...9

3.1 Social decision theories related to aging...9

3.2 Social decisions now ... 11

3.3 Positivity bias ... 11

4. Attention ... 12

4.1 Attention in general ... 12

4.2 Eyetracking/Mousetracking ... 13

5. Present Study ... 14

6. Method... 16

6.1 Overall design ... 16

6.2 Participants ... 16

6.3 Programming ... 17

6.4 Procedure ... 17

6.5 Main variables of interest ... 19

6.6 Attention measures ... 19

6.7 Measures: Self-report ... 20

6.8 Statistical models... 21

7. Results ... 23

7.1 Effect of Age on Choice. ... 23

7.2 Effect of Age on Attention ... 28

7.3 Effect of Attention on Choice ... 32

7.4 Effect of Extraversion on choice ... 36

8. Discussion ... 38

9. Limitations and future research ... 41

10. Conclusion ... 42

References ... 43

Appendix. ... 47


1. Introduction

Throughout life you constantly make social decisions. Humans are immediately social from birth and already from a young age engage in complex social interactions, such as

distinguishing and responding to the emotional states of their caretaker, before even being able to perform basic life functions (Carstensen & Fredrickson, 1990). People are social beings and social connections with others will therefore continue to develop throughout life, but the exact nature of those interactions may change. It has been widely accepted that rates of social interactions decline with age, they become more selective and familiar focused, and this is also supported by research (Carstensen &Fredrickson, 1990). Nonetheless, we will always face decisions that either involve or affect others and it is therefore important to understand how people make social decisions. Additionally, it is important to understand how decisions change and develop over time as social decisions can also have an effect on overall well-being, whether interactions influence cognitive decline, and socio-economic

preferences. But also, more situations or events that happen throughout life, such as change in preference and loss of a close friend or relative, could influence social decisions and a

person’s wellbeing.

Over the past decade, the social possibilities and settings have constantly been changing and evolving, particularly as social interactions have moved online, it is easier to get in contact with strangers across the globe and it has become easier to stay in contact in general. From a younger age people are now able to socially connect with people across the globe and at an older age it is much easier to stay in touch and socially interact compared to a decade ago. Another recent example of how easily our social settings can change is the world-wide corona pandemic, where restrictions were placed on our social interaction, and we had to adapt to a new way of social interaction in a short period of time (Bardach et all.

2021). The term social distancing became the new normal and social interaction had to change. The pandemic raised a lot of questions and concern how this new social setting would influence how children and adolescents would cope and develop their social skills, as well as older people being cut-off from the few ties they still had and how this could

potentially affect their aging and illnesses such as dementia. It is, therefore, important to see how all of this has potentially changed how people make social decisions and whether this has influenced social decision making.

A theory that tries to explain social decision making and age differences is the theory introduced by Carstensen, namely Socioemotional selectivity theory. This is a life-span theory of social motivation where time perception, and especially perceived endings and time limits, plays a central role in preferences for social partners and prioritization of social goals (Carstensen, Isaacowitz & Charles, 1999). The theory suggests that the older we get, the more we become aware that time is short and the more we perceive an ending approaching.

As a result, we prioritise social interactions where emotional regulation is the goal, shifting our social goals from pursuing information to maintaining relations that help regulate emotional states (Carstensen, Isaacowitz & Charles, 1999; Carstensen, 2006). Since the theory was first introduced over two decades ago, a lot has changed in how people socially interact and the social possibilities that we now have, and it is important to see whether this theory is still relevant.

An aspect that has been proven to play a role in decision making in general and influence what decisions are made is attention (Mormann, Towal & Koch, 2013; Lim,

O’Doherty & Rangel, 2011; Krajbich, Camerer & Rangel, 2012). Multiple studies have found that your attention indicates what decision you are going to make and influences the process of decision making (Lim, O’Doherty & Rangel, 2011; Krajbich, Camerer & Rangel, 2012).

Even though this is well known, it has not necessarily been applied to scenarios of social decision making. In addition, it has also been shown that age influences attention such that


when you get older the top-down process, that is the process of using complex information to shape lower-level processes, is negatively affected (Mather & Carstensen, 2005; Lighthall, 2020; Gilbert & Sigman, 2007). More recently, evidence suggests that time horizons and the perceived endings and time limits that are discussed in socioemotional selectivity theory are associated with attention in the context of aging (Charles et al., 2003; Mather & Carstensen, 2003). It is, therefore, also important to see what role attention plays in social decisions for the different generations and how this supports or contradicts socioemotional selectivity theory. Attention has the potential to distinguish different theories regarding social decisions.

The research questions for this thesis therefore are:

1) How do social decisions regarding who to spend time with change over the lifespan?

2) Does attention explain these age differences or individual differences generally? Do age and/or attention have an effect on how social decisions are made?

Contrary to the previous literature, this research found in one of the time-limited scenarios that older adults significantly choose the novel option and younger adults choose the familiar option. Moreover, age itself did not significantly influence attention throughout the

experiment. However, attention towards specific boxes in the person open ended scenario and in the additional information time-limited scenario had a significant effect on the choice of the participant.


2. The ageing brain

It has been established that ageing causes changes in physical aspects of the brain which leads to deterioration of brain function and decline in cognitive abilities (Larkin &Knutson 2015; Esiri 2007, Peters 2006, Bruke & Barnes, 2006). Ageing and physical changes to the brain has also shown to lead to less negative changes, such as positivity bias, which might be beneficial for well-being and could impact decision making (Labouvie-Vief et al., 2010;

Cacioppo, et al., 2011) However, there are still a lot of mixed results as to which specific brain areas are most affected by age and to what extent. Most studies are cross-sectional and focused on age-differences and not necessarily longitudinal studies focused on age changes, therefore, have not seen exactly how the same brain has changed (Fjell &Walhovd, 2010).

Additional to the physical changes, the interaction of age and other contextual factors also influence how the brain is affected (Fjell and Walhovd, 2010). It is also the case that

everyone’s brain ages differently and some brains age better than others (Esiri, 2007). There is no well-accepted comprehensive theory with regards to the aging of the brain and different areas of the brain that have received focus (Gray and Barnes, 2015)., Nonetheless, it is possible to move towards a consensus on the effect of aging on the brain by creating a bridge between the different results and unifying them, as this research is trying to do with attention and social decisions. Nevertheless, different findings regarding how age influences and affects the brain will be discussed.

2.1 Volume

As we age our brain starts to decrease in overall volume and weight (Esiri, 2007; Anderton, 2002; Fjell and Walhovd, 2010) with some areas, such as the cerebral cortex, declining linearly from early in life while most others only start to decline in volume after having peaked in middle adulthood (Fjell and Alhovd, 2010). When we look at the brain, we know that it consists of both grey and white matter, with brain tissue consisting of approximately 48% grey matter (Albert, 1997). Both grey and white matter have been proven to be negatively affected by ageing. The grey matter is important for processing of information while white matter is important for the communication between the different areas of the brain and the rest of the body. It is estimated that brain volume reductions will increase every year and that at the age of 30-50 the reductions in brain volume are around 0.1-0.2%

yearly and around the age of 70 will reach 0.3-0.5% every year (Esiri, 2007). One would expect the shrinkage of the brain to mainly be because of neuronal loss and death, however this is only related to a minor extent (Peters, 2006; Fjell and Walhovd, 2010). In addition to neuronal death, shrinkage of neurons and reduction in synapse density in the dendritic spines in the cerebral have shown to most likely account for reductions in the grey matter of the brain (Fjell and Wlahovd, 2010).

2.2 Grey matter

When looking at grey matter (GM) specifically we see a negative correlation of GM volume with age in most cortical regions (Giorgio, et all. 2010; Fjell and Walhovd, 2010). The

reduction of GM in the brain already starts early in life (Fjell and Walhovd, 2010). Especially the frontal cortex, as well as some parietal and occipital regions seem to be most affected negatively by age as can be seen in the figure 1 from the research done by Giorgio et all (2010). When comparing young and middle-aged adults specifically there is clear reduction in the GM volume in the frontal lobe, including the caudate nucleus and to a lesser extent in the temporal lobe (Giorgio, et all., 2010). The volume reductions in GM that Giorgio et all (2010) found are more widespread compared to some previous reports that reported specific structures such as amygdala and hippocampus (Good et al., 2001). The reduction in the


frontal cortex, amygdala and hippocampus could have implications for social decision making. The reduction could negatively influence functions of these areas that are part of social decision making, such as information processing, decision making in general and emotion. When specifically looking at volume reduction in GM there are a number of potential cellular processes that could explain this (Giorgio, 2010). In early adulthood elimination of synapses and neurons contribute to volume reduction while in middle and late neurons shrink and get smaller and more neuronal loss occurs (Giorgio et al, 2010; Ge et al., 2002).

Figure 1. From the research done by Giorgio et all. (2010) that shows a Voxel-based morphometry (VBM) style analysis of how GM changes with age. (A) The blue-coloured voxels indicate the regions that have

demonstrated a significant negative correlation between GM volume and age. (B) The plot illustrates the relationship between mean GM volume and age across all significant voxels.

The blue triangles represent male subjects, and the orange triangles represent female subjects.

2.3 White matter

When looking at white matter (WM) it seems that the loss in volume in these areas is greater and more rapid than that of GM loss and that it ultimately exceeds it (Fjell and Walhovd, 2010). The most significant linear correlation of WM volume and age that Giorgio et all.

(2010) found was in the right hemisphere of the prefrontal cortex, connections between the thalamus and frontal lobe, the cerebellum, connections between the putamen and claustrum and the anterior midbrain as can be seen in figure 2. It becomes clear that age has a negative effect on connections and communications between areas that are important for decision making, emotional processing and social interaction (Nummenmaa, et al., 2012) The WM volumes relationship with age follows a nonlinear inverted “U-shape” (Ge et al., 2002; Sowel et al., 2003). Giorgio et al. (2010) found that myelinated fibres are affected by pathogenic mechanisms that are known to underly the ageing process in the same way.

Figure 2. From the research conducted by Giorgio et all. (2010) that shows the VBM- style analysis of WM changes with age.

(A) The blue coloured voxels show the significant linear decrease of WM volume with age and the green coloured voxels show the non-linear relationship between WM volume and age. (B,C) The plots illustrate the relationship between mean WM volume and age across all voxels showing both a significant linear and nonlinear relationship with age. The blue triangles represent male subjects, and the orange triangles represent female subjects.


One negative aspect of the study that Giorgio et al. (2010) conducted is that their participants consisted of mainly young adults and that there was a lower number of

participants in the middle and older adults’ subgroups. In addition, it was all participants of the same undisclosed region and environmental factors have also proven to have an effect on the brain so in order to have a more accurate representation of the entire population more subjects from different areas and more older subjects should be recruited. It is also difficult to determine what the exact change in brain structure was as they are cross-sectional studies comparing different brains and not longitudinal studies where they look at one brain and how it changed over time. In the future it would be interesting to keep track of the same group of brains and see how they change exactly over time.

2.4 Hippocampus

Aging has often been shown to cause learning and memory problems many of which have been associated with hippocampal damage (Driscoll, et al., 2003). The hippocampus is the brain structure that is crucial for episodic memories, which is the memory for personally experienced events that are set in a spatio-temporal context, and is also crucial for storage and retrieval of information and in the feedback control of the response to stress (Gray and Barnes, 2015; Burgess et al., 2002; Miller and O’Callaghan, 2005). Damage to the

hippocampus has proven to result in memory defects (Gray and Barnes, 2015). The hippocampus is also particularly vulnerable to the aging process and over the age range of 30-90 has shown the largest reduction of its volume around 35% (Burke and Barnes, 2006;

Anderton, 2002; Fjell and Walhovd, 2010). Therefore, it is not surprising that information processing in this brain region is negatively affected by aging and declines over time and, similarly to damage to this area, show memory defects increase during normal aging (Burke and Barnes, 2006; Gray and Barnes, 2015). Driscoll et al. (2003) conducted a research where they looked at the hippocampus of young adults and elderly and how they performed on hippocampus dependent tasks including the virtual Morris water task (VMWT) and the transverse patterning discrimination task (TPDT) which are both navigation and pattern discrimination tasks. They also found a negative relationship with ageing in the performance of the participants and a significant decrease in hippocampal volume. Therefore, they found that larger total hippocampal volumes are associated with better performance on

hippocampus dependent tasks when looking at the individual differences. Age and reduction in volume, therefore, has a negative relation to how the hippocampus performs. The

problematic aspect of this research is how they only compare young to old and therefore cannot see how the hippocampus has developed and how performance decreases over time.

In addition, the participants were all from the same region and there was only a small pool of 16 young and 16 elderly participants. Regarding how the hippocampus is affected by aging and volume loss there are mixed results. Some claim that the hippocampus is affected by means of neuronal loss, decrease in synaptic connections and atrophy in this region

(Anderton, 2002; Fjell and Walhovd, 2010; West, 1993; Driscoll, et al., 2003), while others claim that neuronal loss is minimal and remain the same in the hippocampus during normal aging (Burke and Barnes, 2006; Albert, 1997; Miller and O’Callaghan, 2005). Reported neuronal loss in the hippocampus would reduce the information processing capacity whereas a loss of synaptic connections would negatively affect memory and learning (West, 1993;

Mayford, Siegelbaum & Kandel, 2012). It is difficult to interpret percentage change in the hippocampus as it is a non-linear age relationship and seems to depend on the individual participant as all brains are affected differently (Fjell and Walhovd, 2010). Therefore, it is difficult to determine the exact effect of aging on the hippocampus, however it is a fact that deficits accompanying normal aging are caused by changes in the hippocampus and the hippocampus is vulnerable to the effects of aging (Driscoll, et al., 2003).


2.5 Prefrontal cortex

Besides the hippocampus, the frontal cortex, and especially the prefrontal cortex, have shown to also be one of the most affected regions of the brain by ageing (Peters, 2006; Anderton, 2002; Yanker, Tu and Loerch, 2008; Fjell and Walhovd, 2010). The prefrontal cortex is involved in a wide variety of executive functions and plays an important role in cognitive control (Miller and Cohen, 2001). The prefrontal cortex is a collection of neocortical areas and communicates with motor systems, cortical sensory systems, and many subcortical structures (Miller and Cohen, 2001). It is involved in learning, memory, social processing, attention and decision making including social decision making (Miller and Cohen, 2001;

Tisserand and Jolles, 2003). A prominent theory on the deterioration of the frontal brain regions is the frontal ageing hypothesis (Lighthall, 2020). This theory suggests that frontal brain regions, including the ventromedial prefrontal cortex that is associated with social decision-making, go through greater deterioration due to aging and cause greater declines in the functions it supports (Lighthall, 2020). Ageing has seen to have an effect on the prefrontal cortical volume, prominent decline due to atrophy, there are increased white matter lesions, the white matter density is negatively influenced and an increase in perseverative behaviour (Peters, 2006; Yanker, Tu and Loerch, 2008 ). Furthermore, reduced activation in the prefrontal cortex has been identified when performing executive processing tasks (Yanker, Tu and Loerch, 2008). There was specifically a reduced activation in the dorsolateral PFC, which is involved with working memory, in older adults compared to younger adults when performing executive and memory tasks (Tisserand and Jolles, 2003). This is in accordance with many neuropsychological studies that discovered that executive functions, that depend heavily on the frontal neural circuits, are most negatively affected and vulnerable to the effects of ageing (Fjell and Walhovd, 2008; Tisserand and Jolles, 2003). There is a clear decline in working memory, explicit memory, attention and learning with age (Tisserand and Jolles, 2003).

As has become clear aging has a negative effect on the brain as overall brain volume decreases, GM and WM are both negatively affected and decrease in volume and specific brain regions such as the hippocampus and prefrontal cortex are negatively affected in their activation and functioning. These areas are associated with social processing and decision making and therefore could influence how social decisions are affected over time. This study is not able to study the brain itself, however, these changes in the brain could explain the underlying mechanism of how age effects social decisions. Therefore, understanding these differences matters if we want to understand societal issues associated with ageing.

3. Social preferences over the life span

3.1 Social decision theories related to aging

There are multiple theories as to why social decisions decline with age, such as fewer social contacts and social interactions in general. One such theory is activity theory that suggests that the reduction in social contact is reflecting barriers to interaction such as deaths of loved ones and limitations due to health problems (Maddox 1963; Löckenhoff & Carstensen 2004) Another theory, that is in fact in opposition to the activity theory, is the disengagement theory. This alternative model of declining social activity claims that reductions in social interaction in fact reflects a natural process. Older people choose to decrease their

interactions as a preparation for their own death (Cumming & Henry 1961; Fredrickson

&Carstensen, 1990). As an alternative to both the activity and disengagement theory Carsten (1987,1989,1990) introduced selectivity theory. Selectivity theory proposes that older people try to minimize negative effects, such as negative experiences and the risks involved, and


want to optimize the experience of positive effects, such as positive experiences (Fredrickson

& Carstensen, 1990). They do this by reducing levels of social interaction, being more selective of who they spend time with, focussing on familiar relationships and higher quality interaction (Fredrickson & Carstensen, 1990). A demand to conserve energy and promote positive emotions increases with age and as a result, criteria for choosing social partners changes (Fredrickson & Carstensen, 1990). For example, it is possible that older adults judge familiar partners as more desirable and novel partners as less desirable as familiar partners will lead to more predictable and possible positive affects (Fredrickson & Carstensen, 1990).

Additionally, the older adults also have a limit of future possibilities with novel social partners as developing long-term friendships with new people is unlikely due to the limited time they have left (Fredrickson & Carstensen, 1990). Carstensen further builds on this theory by emphasizing that the selection and pursuit of social goals is also inevitably linked to the perception of time. As people get older, they become more aware that time is “running out” and making the “right choice” and not wasting time becomes important (Carstensen, Isaacowitz & Charles, 1999). These arguments are further developed in socioemotional selectivity theory (SST) (Carstensen 1991, 1993, 1995, 1998, 1999). This is a similar life- span theory that states that constraints on time horizons shift motivational priorities regarding social decisions in such a way that emotion regulation becomes the most important goal (Carstensen, 2006). Previous research supporting SST has found that different ages do prioritize different goals when it comes to social decisions (Carstensen, 2006). As people grow older and start to perceive time as limited, they start to attach greater importance to goals from which they derive emotional meaning and attach less importance to goals of obtaining knowledge and expanding their horizons (Carstensen, Isaacowitz & Charles, 1999;

Carstensen, 2006). SST suggests that the goal of knowledge acquisition starts high in the earlier years of our lives and over time declines as knowledge is accumulating and the concept of future becomes shorter (Carstensen, Isaacowitz, Charles, 1999). These knowledge and emotion trajectories can be seen in figure 3. SST further states that as a function of perceived time there are two categories of goals shifts that direct social behaviour: goals concerning the acquisition of knowledge and information; and goals concerning the

regulation of emotional states (Carstensen, 2006). Most of our social behaviour is motivated by the pursuit of information and knowledge (Carstensen, Isaacowitz & Charles, 1999;

Sharot & Sunstein, 2020). Knowledge acquisition through social contact is needed to

understand the social climate, to get to know our own preferences and to develop non-social skills (Castensen, Isaacowitz & Charles, 1999). The category where knowledge acquisition is the goal therefore refers to acquisitive behaviour that is focused on learning about the social and physical world (Carstensen, Isaacowitz & Charles, 1999; Sharot & Sunstein, 2020). The category of emotional motives has to do with regulating emotional states through contact with others as our emotions are shaped in a social context (Carstensen, Isaacowitz & Charles, 1999). This category also includes the desire for emotional intimacy, establishing a feeling of social connection and the desire to find the meaning in life (Carstensen, Isaacowitz &

Charles, 1999). According to SST both goals together form a constellation that motivates our social interactions, and when they compete in a social situation, they are weighed against each other and action is taken accordingly (Carstensen, Isaacowitz & Charles, 1999). An aspect that influences which goal is prioritized is that influencing the perception of time.

When time is perceived as open-ended goals regarding gathering information, experiencing novelty, and expanding knowledge are prioritized (Carstensen, 2006). While when time is perceived as limited and constrained the most important goals will be those that can be achieved in the short-term and that emphasize feeling states and emotional states to increase psychological well-being (Carstensen, 2006).


Figure 3. Knowledge and emotion trajectory graph from the Carstensen, Isaacowitz and Charles (1999) paper.

3.2 Social decisions now

SST was first introduced by Carstensen et al. in 1990 and most of the research she and her colleagues have performed is over a decade old. Since then, a lot has changed in how people socially interact, and online social platforms are used to socially connect by people of different ages (Chiarelli & Bastistoni, 2021). Chiarelli & Bastistoni (2021) looked whether SST was also accurate in older people’s activity on Facebook. Their results, that older adults had smaller and closer networks online, partially supported the theoretical logic of SST. This, therefore, does show that the theory of SST extends to an online platform. However, in their own research and results they only focused on the activity of older adults with no comparison to younger adults and the participants were mainly female and from Brazil. Technology has provided opportunities to stay in touch which was especially important during the COVID restrictions where social distancing was encouraged and people had to live in social isolation (Bardach, et. all., 2021). However, social contact was significantly reduced, and the

guidelines experienced were rather isolating. Therefore, this could potentially influence social decisions about who to interact with. Older adults may miss the physical contact of their loved ones while younger adults may miss the social contact of new potential social interactions which would be in line with SST. However, the pandemic could have also had the opposite effect. Older adults might realise they miss the new social interactions with strangers and younger adults may value their close contacts more than before. That is why especially now it is interesting and important to see and understand how social decisions have been affected.

3.3 Positivity bias

Older adults exhibit certain decision biases more strongly with age, and one well documented bias is the positivity bias (Daniels and Zlatev, 2019). This bias is when we are more drawn towards decisions that have more positive outcomes and have more positively valanced choice sets and information (Daniel and Zlatev, 2019; Levin, Fiedler, and Weber, 2021). It turns out that there is an age-related increase in the preference for positive information over negative information (Reed & Chan, 2014). Older adults tend to remember, process, and attend information with a positive valence to a greater extent versus negative information compared to younger adults (Carstensen & Mickels, 2005; Levin, Fiedler, and Weber, 2021;

Daniel & Zlatev, 2019). This age by valence interaction is known as the positivity effect (Levin, Fiedler, and Weber, 2021). This age-related positivity effect has been linked to age- related cognitive and neural deficits. Labouvie-Vief et al. (2010) proposed the Dynamic Integration Theory in which they state that age-related cognitive declines push older adults to


have an automatic preference to process positive information over negative information because it is easier to process. In addition, the aging brain model states that the positivity effect has to do with degeneration of the amygdala, that is involved in decision making and emotion regulation, that in turn is inhibiting neural and affective responses to negative information (Cacioppo, et al. 2011). This positivity bias also helps with emotion regulation, and thus can be linked to the motivations in SST that influence social decisions (Pruzan &

Isaacowitz, 2006; Levin, Fiedler and Weber, 2021). Older adults have been found to be more sensitive to emotional cues when making social choices and the positivity bias and effect could therefore guide their choices (Levin, Fiedler and Weber, 2021; Charles and Carstensen, 2010; Hess, 2005).

4. Attention

4.1 Attention in general

Attention, defined as a selectivity in perception, is found to be one of the key factors that influences choice (Orquin and Mueller Loose, 2013; Mormann, Towal & Koch, 2013). For example, when looking at consumer choice behaviour, when a consumer repeatedly allocates attention to a product, the likelihood of purchasing that product increases (Janiszewski, Kuo

& Tavassoli, 2013). By examining consumers’ attention traces, which is defined as a

sequence of eye fixations, we can see and examine the decision strategies, consideration sets and alternative evaluations (Mormann et al., 2020). Attentional processes in general are known to both be influenced by top-down goal directness and bottom-up environmental cues (Janiszewski, Kuo & Tavassoli, 2013). They are therefore known a source of preference formation that are driven by preference, top-down processing, and influenced by preference, bottom-up processing, in an interactive process. Attentional processes accordingly help select information that is deemed necessary and important to make subsequent judgements and eventual decisions (Janiszewski, Kuo & Tavassoli, 2013). Earlier theories such as bounded rationality models and strong rational models suggested that attention only passively served the decision process, however, more recent models have proven that attention in fact plays and active role in the decision-making process (Krajbich, Armel and Rangel, 2010). One of these models is the drift diffusion model that states that decisions are based on evidence that has been accumulated during fixations (Krajbich, Armel and Rangel, 2010). Therefore, studying attention helps us better understand actual consumer choice, decision-making activities and how decisions are made (Mormann, et al., 2020).

The time horizon that plays an essential role in SST has also been linked to both attention and memory in relation to aging and adult development (Mather & Carstensen, 2003; Charles et al. 2003). Mather & Carstensen (2003) found that older adults exhibit an attentional bias towards positive and away from negative information. Furthermore, it was revealed that older adults attend more to positive information when the attentional processes are more influenced by top-down control (Marther & Carstensen, 2005). Top-down processes have been proven to have greater structural and functional decline in normal aging and is in line with the frontal aging hypothesis (Lighthall, 2020). It seems motivational difference over the lifespan are likely related to structural differences in ages and these structural-functional changes influence how people make social decisions and where their attention goes.

Additionally, the fact that older adults are more drawn to positive information indicates that the positivity bias and effect are present when making social decisions. However, the combination of social decisions and attention has not been studied extensively. Pruzan &

Isaacowitz (2006) investigated the attentional effects of anticipated social endings between students and based it on the SST theory. They used an eye-tracker and showed the

participants different faces expressing different emotions. Even within this small age


difference they found that the anticipated ending did influence the participants attention and senior students avoided negative images, that contained negative facial expressions (Pruzan

& Isaacowitz, 2006).

4.2 Eyetracking/Mousetracking

Multiple studies have shown a strong link between attention, specifically visual attention, and eye movement (Orquin and Loose, 2013). While in the past researchers asked participants to think out loud and describe where they are looking and what they are doing, eye-tracking gives a more precise and objective measurement as to where the attention goes and what information is important when making a decision (Diamasbi, 2014). Eye-movement research has therefore become of great interest across many domains and in the fields related to neuroscience, ergonomics, marketing, design, and web experience (Richardson and Spivey, 2004; Diamasbi, 2014). Eye-tracking gives researchers the possibility to see either the product, website, or experiment from a user’s point of view and provides information regarding the user experience and how visual attention is distributed (Diamasbi, 2014;

Blascheck et al., 2014). Eye-tracking itself captures the focus of a viewer’s gaze, that is an externally observable indicator of visual attention, on a stimulus at a given time (Diamasbi, 2017). It is not 100% correlated, however, is often a good measure that relies on the

assumption that people look at what they are thinking about and paying attention to. Eye- tracking systems used today capture eye movement in an unobtrusive way through video- based corneal reflection (Duchowski, 2007; Diamasbi, 2014). Information that is obtained from eye tracking is often about fixations, that is defined as stable gazes with a minimum time threshold (Diamasbi, 2014). Suitable fixation thresholds range between 60 and 100 miliseconds depending on the complexity of the scenery and information present (Diasmasbi, 2014). Combining all fixation points gives the fixation pattern which provides invaluable information about the spatial distribution and order of attention on the stimuli, (Diamasbi, 2014). Besides eye trackers creating fixation patterns it is also possible to focus on specific regions or areas, also known as areas of interest (AOIs) within which measures such as fixation duration, fixation frequency, fixation timing and percentage of viewers can be calculated (Blascheck et al., 2014; Diamasbi, 2014). As can be expected a large amount of data is generated during eye tracking experiments, which could be very time consuming in the analysis and could be considered as a small disadvantage (Blascheck et al., 2014).

In addition to eye-tracking, mouse-tracking is another method that can capture attention and how people make decisions. As the name already suggests mouse-tracking is when computer-mouse movements and trajectory made by participants, often between 2 options, are measured and tracked while choosing between choices (Stillman, Shen and Ferguson, 2018; Kieslich, et all., 2020). Mouse-tracking has become an emerging tool that, like eye-tracking, offers data-rich, accessible and a real-time view into how people make decisions and their cognitive processes (Stillman, Shen and Ferguson, 2018; Kieslich et all., 2020). It is an effective tool and can approximate gaze, however even though it is a good alternative it cannot be used as a full substitution for eye-tracking (Huang, White and Busher, 2012). It has already been used in fields regarding action control, perception, memory, value- based decision-making, social cognition, and many other disciplines (Kieslich, et all., 2020).

Mouse-tracking has also been proven to be an effective manner of studying decision conflict and there are two major advantages of using mouse-tracking to examine theoretical

predictions regarding decision conflict. The first being that it provides a real-time window into the temporal development and resolution of how a conflict is resolved and secondly it has the potential to show more precisely the amount of conflict present during any type of decision (Stillman, Shen and Ferguson, 2018; Kieslich, et all., 2020). In this sense, mouse- tracking has been used to differentiate between dual system models and dynamic models


(Freeman & Dale, 2013). Similar to mouse-tracking softwares that measure the mouse trajectory, you also have other mouse tracking systems such as Mouselabweb. In

Mouselabweb you do not record the mouse trajectory between two options but instead have different areas of interest where you can hover your mouse to gather information (Willemsen

& Johnson, 2019). The information is often hidden beneath overlaying boxes therefore encouraging people to hover over these boxes to access the information. This allows you to record the frequency, duration and sequence of the mouse entering the different boxes and options presented. Subsequently, it gives you the opportunity to create more complex setups (Willemsen & Johnson, 2019). It offers the same advantages as mouse-tracking and has additional assets such as that it can be used outside of a lab and online. Another advantage is that compared to an eye-tracking experiment all you need in a mouse tracking experiment is a computer and its accompanying hardware (Hehman, Stoller & Freeman, 2015). One aspect that is a constraint is that you are only able to access a single piece of information at once.

5. Present Study

The present study will attempt to assess how age and the attention of the participants

potentially influence their social decisions across the lifespan. In contrast to the research done by Pruzan and Isaacowitz (2006) the participants will be of different age groups and instead of eye-tracking, Mouselabweb will be used. It is therefore possible to apply the results to people of all ages and an online method will be used to assess attention in order to make the study more scalable. In addition, the experiment will be based on the card game that

Fredrickson & Carstensen (1990) used, and participants will be asked to decide who they would like to spend time with in open-ended and time-limited scenarios. The time-limited scenario has been changed from moving across country to a more recent and relevant time- limiting scenario regarding quarantine, so it was easier to grasp for participants and to remind them of the lack of social contact and how we suddenly had to socially distance (Bardach et all., 2020). In addition to the information of different people to decide between, an extra element has been added to the experiment. While in previous research the aspect of

‘knowledge’ that could be gained was more implicit in the type of person and was not emphasised, in this experiment participants will also be given extra information about the people they can interact with regarding learning value. For each option of who to interact with there will be some additional information each indicating a level of informative value of the person the participant can spend time with, such as what the person has experienced lately. The different people they can decide between, and additional information will be hidden under boxes and to reveal this information participants will have to hover their mouse over the boxes. By having all the relevant information under the boxes, the attention of each participant will be tracked to see what they look at before making a social decision.

5.1 Hypotheses

As stated above the questions for this study are:

1) How do social decisions regarding who to spend time with change over the lifespan?

2) Does attention explain these age differences or individual differences generally? Do age and/or attention have an effect on how social decisions are made?

The first question is based on theories regarding social decision making. Especially the SST that states that older adults’ social decisions become more focussed on those with positive emotional value and therefore often choose for people considered familiar in both open and time limited scenarios (Castensen, Isaacowitz & Charles, 1999). As we live in a current situation where social interaction has been limited, and has been for the past year, it is not


strange to expect that this will have influenced how people of different ages want to spend their time and the social decisions that they make (Bardach, et al., 2021). Based on the SST theory and current social climate the following hypotheses have been formulated:

H1: Older adults will have an even stronger emotional goal when making social decisions in an open-ended scenario compared to younger adults.

H2: Both older and younger adults will experience an even stronger emotional goal when making social decisions in a time-limited scenario.

The second question is to try to see if attention could potentially bridge the literature and explain how age influences social decisions and whether attention itself plays a role in how social decisions are made. It is clear that attention plays a role in decision making and age negatively influences brain areas, such as the frontal cortex, that are involved with both attention and social decisions (Janiszewski, Kuo & Tavassoli, 2013; Lighthall, 2020; Miller and Cohen, 2001; Tisserand and Jolles, 2003). According to the positivity bias older adult’s attention is drawn towards positive information compared to younger adults (Marther &

Carstensen, 2003). Moreover, as stated in the SST older adults go for the familiar option most because they want to regulate their emotions in a positive way while younger adults focus on expanding their knowledge and horizons (Carstensen, Isaacowitz & Charles, 1999). By combining these theories, the following hypotheses were created:

H3: Older participants attention will be more towards familiar and therefore emotionally positive choices compared to younger adults.

H4: Older participants will only focus on the information regarding their relationship to the person while younger participants will look more at the learning value of the person.

H5: Attention towards the familiar boxes will lead to the familiar choice and attention towards the novel boxes will lead to the novel choice.


6. Method

6.1 Overall design

The main goal of the experiment is to see what role attention plays in social decision making across the adult lifespan. The experiment was based on the card experiment by Fredrickson &

Carstensen (1990) in which people were asked to make the social decision of with whom they would like to spend time. Identical to that experiment, participants were first introduced to the scenario which was either open-ended or time limited. To make it relevant to the environment, with the current COVID pandemic, the time limited scenario was linked to going into quarantine, so participants were able to relate and recognize the situation they were in. To investigate attention in social decision making the overall experiment was coded in MouselabWeb, which is a mouse tracking platform that can be used online (Willemsen &

Johnson, 2019). In the first half participants were then introduced to three different categories of people they could choose between. One familiar person, one categorized as novel-familiar, and the final person to choose from was a novel person. The different people and categories the participants were exposed to are the same people and categories used in the Carstensen experiment. In the second half an extra element was added to the design. Besides the different people to choose between being hidden under boxes, additional boxes containing extra

information about that person were added. The information within these boxes indicated something the participant could potentially talk about with the person and therefore learn from them. The knowledge aspect and learning value that each person was implicitly associated with has now been given its own box and therefore the attention can be tracked whether it goes more towards the person/familiarity information or the learning-value information and whether this changes the social decision.

6.2 Participants

The participants in this study were 76 adults that are divided into 6 different age groups. The first three age groups are considered the young adults and the last three age groups are considered the old adults. Within the different groups range 44 participants were part of the young adults and 32 were between part of the old adults. Because in the experiment the participant did not give their exact age, but rather picked the age group they belonged to, we do not have an accurate average age of both groups. However, to give a slight indication of average age of the participants we took the midrange of the different groups as an

approximation. This gave us an average age of 27 (sd.=4.45) for the young adults and 52 (sd.=5.90) for the older adults. Of the participants 32 were male and 45 were female. There were multiple different nationalities and residencies with the main ones being the

Netherlands and the United Kingdom.

Participants were recruited in different ways. Most of the participants were through convenience sampling and included family, friends, and acquaintances. In addition, a post on LinkedIn that was shared multiple times and snowball sampling occurred as the experiment link was distributed by other participants. Participants of all ages and backgrounds were included to have a diverse group of participants, particularly with regards to age. Data was collected from 01-06-2021 until 01-07-2021.

As an incentive for participation, one participant of the experiment would randomly be selected to receive a gift card of 25 euros. Participants could take part by submitting their email at the end of the experiment. The email was not linked to any of the data and therefore preserved anonymity. In addition to the randomly selected participant there was also a chance to win another 25-euro gift card. Participants could win this gift card by being the most accurate at predicting how the different ages made social decisions in the experiment.


6.3 Programming

To be able to conduct the experiment on an online platform the experiment had to be coded with php and json files. The creator of Mouselabweb, Willemsen, already had some pre- coded php and json files that you are able to download and change to make it fitting for your experiment and could be found on GitHub (Willemsen, 2021). These pre-coded files were to make sure that the mouse tracking data would be tracked and stored accordingly. The files were then recoded to resemble a similar set-up as the Fredrikson & Carstensen (1990) card experiment. The files were then uploaded to a web domain brought on Strato called through the program FileZilla. An example of the experiment webpage can be seen in figure 5 and 6. As part of the code that was downloaded from GitHub there were also additional files to help download and partly summarize all the choice and attentional data recorded throughout the experiment of each participant. A threshold of 100ms was chosen when downloading the attentional data as this threshold is close to the recommended 200 ms and provided the most data (Willemsen & Johnson, 2019).

6.4 Procedure

The experiment itself was then divided into three parts.

Figure 4. Procedure of the entire experiment. First part is the Social-Decision making. Second part is the closeness & learning value, motivation, incentive and personality. Third part is the demographics survey.

In the first part, participants completed 20 trials of social decisions, consisting of 2 different scenarios, each consisting of 10 trials. On each trial, participants are presented a scenario and can reveal the three different people they can spend their time with by hovering their mouse over the boxes on their screen. The categories of people they can choose between are: one familiar person, for example a close friend or family member, one categorized as novel- familiar, which entails someone they are acquainted with but not as close as the familiar person, and the final person to choose from was a novel person, that could be someone that they recognize but do not necessarily know such as a famous person. In the first 10 trials, participants are only given the information of whom they could decide between before

making a decision. Of these 10 trials, 5 are with the open-ended scenario, person open-ended, and 5 are with the time-limited scenario, person time-limited. In the final 10 trials additional information regarding those people is added in boxes under the people they could choose between. The additional information is specific to the option of whom to spend time with and could give the participant an indication what they could talk about or learn from that person.

Just like in the previous 10 trials, participants are exposed to both scenarios equally. First to the time-limited scenario, additional info time-limited, followed by the open-ended scenario, additional info open-ended. In each trial they are asked to decide who they would like to spend time with.


Figure 5. Example of the first half of the experiment with the first scenario (open-ended). Here the scenario is open ended, and you have 3 boxes of information regarding the person you can spend time with. The information in the boxes can be accessed by hovering the mouse over the box.

Figure 6. Example of the second half of the experiment with the second scenario (time-limited) Here the scenario is the quarantine time-limited scenario with 6 boxes of information regarding the person you can spend time with and additional information. The information in the boxes can be accessed by hovering the mouse over the box.

The second part of the experiment consists of different self-reporting measures. Participants are first asked to rate how close and familiar they felt to the different people they were presented with throughout the trials on a 7-point scale created by Gächter, Starmer & Tufano (2015). Next, they were asked to rate the learning value they associated with each person also on a 7-point scale. These two measures are asked to see whether the categories given to the different people is accurate and to see whether participants saw that person as more of an emotional choice or knowledgeable choice.

After indicating the closeness and learning value, participants are asked to give a short description on their motivation behind their choices in the trials with 3 boxes and those


with 6 boxes. They are also asked whether they looked more at the person or extra

information when making the choice and whether they considered closeness or learning value as more important.

This is followed by an incentivised part where they are asked how they believe other people of different age groups made decisions in both of the scenarios. They were

specifically asked whether they went for the familiar, novel-familiar and novel people. In addition, they were also asked whether they think the different age groups went more for familiar/closeness or for learning value/information and which one they looked at most in the second scenario. The participant with the most correct answers will receive a gift card of 25- euro as a reward. This incentivised part is to get people thinking about how people of their own group responded in the experiment and also how others act when making a social decision.

The final self-reporting measure was a brief 10 item Big-Five personalities dimensions test by Gosling, Rentfrow & Swann Jr. (2003). This measure consisted of different personality traits where participants had to indicate how much they recognized themselves. This test was to see if the characteristic Extraversion also played a role in the social decisions made.

The third and final part of the experiment was a demographic survey where participants were asked about their personal information but also about their social

interaction before and during the COVID pandemic and about their health status as possible mediators for individual differences in social choice.

6.5 Main variables of interest

Independent variables. Age category is the primary independent variable of interest as we are trying to see whether age influences a person’s social decisions. Participants could choose between 6 different age groups therefore making age an ordinal variable. In addition,

attention to the information in boxes is also an independent variable. Attention was measured in two ways: 1) as the amount of time the participants mouse spent in a box (in milliseconds) and 2) how often they accessed the information in the box, the frequency. Frequency was the main variable used in the statistical analysis as this is more controlled and reliable compared to time.

Dependent variables. After participants have attended to the information available to them and have looked at the different boxes, they are asked to make a social decision of who to spend time with among three options: a familiar person, a novel-familiar person and a novel person. An example of what the different trials and boxes look like can be seen in figure 5 and 6. As the options of novel-familiar and novel are quite close and both different from what is considered the familiar option they have been combined as the novel option for the

analysis. Therefore, the main dependent variable will be choice which consists of two categories, familiar and novel. For each scenario the most common choice is taken for each participant so one dependent variable is used for each scenario.

6.6 Attention measures

Mouse tracking. To be able to measure the attention of the participants while making social decisions, the process tracing tool Mouselabweb was used. It measures information search to study cognitive processes involved in decision making in an online platform. With the help of Mouselabweb, boxes were created that hid important information for the participant that was necessary to make the social decisions. To be able to see the information the participant must hover their mouse over the box to reveal the information in the box. As soon as the mouse


enters the box, the time that the mouse enters the box is registered and similarly when the mouse leaves the box the time is also registered. Therefore, the time in the box is measured and it becomes clear how often the attention of the participant goes to that box of information and for how long. For each trial there were either 3 or 6 boxes and every entry and exit of any of the boxes was recorded as a result tracking the attention of the participant.

6.7 Measures: Self-report

Closeness value. It is relevant to see how and if social relationships had an influence on how social decisions are made. Additionally, the scale is also used to check whether the categories assigned to the different people to choose between were indeed accurate. To be able to

understand and rate participants social relationships with their options of people to spend time with, the ‘Inclusion of the other in the self (IOS)’ closeness scale by Gächter, Starmer &

Tufano (2015) was used. With figure 7 as a reference, participants were asked to rate the closeness to person X on a scale of 1-7. The more overlap there was between ‘You’ and ‘X’

the closer you considered that person. Participants were asked to indicate the closeness for all the different people they came across throughout the experiment.

Figure 7. Inclusion of the other in the self (IOS) closeness scale by Gächter, Starmer &

Tufano (2015). The more ‘You’ and ‘X’

overlap the closer you consider yourself to person ‘X’.

Learning Value. It is also important to measure the learning value that each participant associated to that person to indicate how much knowledge they believe they could gain from interacting with that person. This can help confirm whether the predictions of SST hold and confirm the accuracy of the different categories assigned. Participants were asked to rate each person on a scale of 1-7 as to what they believed the learning value was and how much new information and knowledge they believed could be gained.

Motivation. To understand the motivation behind the social decisions made in the experiment participants were asked to give a short open-ended explanation of how they made their decision and what the motivation was behind their choices. Furthermore, participants were asked whether the motivation behind the choices differed when they were presented with more information in comparison to only being presented with the people they could choose between. These questions were left open so participants could explain their thought process in their own words. This was then followed by a specific question whether the participant looked more at the person or at the extra information provided or both equally when making social decisions in the second part of the experiment. This can confirm how aware

participants were of any variations in their attention, since it was directly tracked during the decisions.


Personality. To account for the possibility of personality playing a role in how the social decisions are made in the experiment, a brief measure by Gosling, Rentfrow & Swann Jr.

(2003) of the Big-Five personalities dimensions test was used. The brief measure consisted out of 10 items regarding different personality traits that may or may not apply to the

participant. The participant was then asked to rate on a scale of 7 how much they agreed that this is a trait that they have ranging from disagree strongly to agree strongly. The personality most relevant to this study was that of extraversion as this can influence how you are in a social setting and how much you seek out interaction with others (Swickert, Rosentreter, Hitner & Mushrush, 2002). Therefore, only two items from this measure will be considered, Extraversion and Reserved. The 1-7 scale of reserved was reversed and had to be recoded.

6.8 Statistical models

In order to answer the research question and the hypothesis three different statistical tests were performed.

(1) The effect of Age on Choice (2) The effect of Age on Attention (3) The effect of Attention on Choice

Before the data could be used and analysed, it was inspected and adjusted to make it suitable for all three statistical tests.

First the variable AgeGroup was created. In the experiment survey participants were able to choose between 6 different age categories ranging from <20, 20-30, 31-40, 41-50, 51-60 and

>60. AgeGroup consists of two categories namely Young adults (=1), under the age of 40 and Old adults (=2), ages 41 and over to simplify the age variable.

Furthermore, choices made by participants for each trial were categorized into F, which contained the familiar choices and is the baseline, and N, which contained the novel familiar and novel choices and is alternative category. For each of the different scenarios the most picked choice was turned into a variable as the main choice of the participant for those scenarios. The people open-ended (Choice1_5), the people time-limited (Choice6_10), additional info time-limited (Choice11_15) and the additional info open-ended

(Choice16_20). Moreover, for each trial attentional data was gathered and similar to the Choice variables was combined to have a total amount for each of the different scenarios.

This was the case for all the familiar options and novel options and for both the frequency and time attentional data.

(1) The effect of Age on Choice

In this statistical test, Choice is the dependent variable and Age is the independent variable.

As both variables are categorical variables a chi-square test to test for association between the categories is performed. Both categories consist of two categories with Choice consisting of familiar and novel and AgeGroup consisting of young and old therefore creating a 2x2 table.

If the test is significant this would indicate that there is an association between the choice made and the age of the participant. In order to see the direction of the choice and whether this has to do with age and other control variables, logistic regression, was also performed.

For choice the baseline was the familiar choice, and the alternative was the novel choices. For age, young adults were the baseline and older adults was the alternative. Therefore, the equation for the logistic regression is:

(1) ln ( 𝑝

(1 − 𝑝)) = 𝛽0+ 𝛽1∗ 𝐴𝑔𝑒𝐺𝑟𝑜𝑢𝑝𝑖

Where p in the equation is the expected probability of the choice being familiar or novel and the beta is the shift due to AgeGroup.


(2) The effect of Age on Attention

To see what the effect of AgeGroup on Attention is in this experiment the Mann-Whitney U statistic test, which is the equivalent to the rank-sum test, will be performed. This is because the predictor variable, age, is categorical and the outcome variable, attention, is continuous.

Usually an independent t-test would be performed however because the assumptions of linear model are not met, and the predictor variable has 2 independent categories, the best test is the nonparametric 2 independent sample test which includes the Mann-Whitney U test.

(3) The effect of Attention on Choice

In this statistical test Choice is the dependent variable and Attention is the independent variable. For the variable attention there are two types of attentional data: the frequency that the boxes are opened, and the time spent in the different boxes. Both the frequency and the time spent will be considered as the attention of the participant and will therefore be tested separately on how they affect the choices made. However, the main variable for attention in this study will be the frequency. As the outcome variable is categorical and the predictor variable is continuous the statistical test will be a logistic regression. The equation for the logistic regression therefore is:

(2) ln ( 𝑝

(1 − 𝑝)) = 𝛽0+ 𝛽1∗ 𝐴𝑡𝑡𝑒𝑛𝑡𝑖𝑜𝑛𝐹𝑖 + 𝛽2∗ 𝐴𝑡𝑡𝑒𝑛𝑡𝑖𝑜𝑛𝑁𝑖

Where p in equation 2 is the expected probability that Attention to familiar or attention to novel choices and information changes the choice from the baseline to the alternative.


7. Results

A total of 76 participants completed the experiment until the end. All 76 were used for the statistical tests regarding Age and Choice; however, only between 50-59, depending on the scenario, were included in the analysis regarding Attention. This was due to the lost data regarding attention that occurred due to the threshold of 100 ms or as the summarized data of participants that used a mobile device for the experiment was not able to be processed for downloading. Of the 76 participants, 44 were young adults and 32 were old adults. For the person open ended scenario there was attentional data for 56 participants of which 31 were young adults and 25 were old adults. For the person time-limited scenario of the 59

participants, 32 were young adults and 27 were old adults. For the additional information time-limited scenario there was attentional data for 50 participants of which 30 were young adults and 20 were old adults. Finally for the additional information open-ended there were a total of 52 participants of which 30 were young adults and 22 were old adults. The uneven distribution between the age groups was due to the way participants were recruited and were willing to participate as it was mostly voluntary.

The results of the additional measures such as closeness and learning value were merely to make sure that the people participants could choose between were in fact in the right category of familiar and novel. The complete table of this can be found in Appendix A.

Overall, the results of this were as expected for closeness. For closeness the mean of familiar was 4.277(sd.=0.844) and mean of novel was 2.378(sd.=0.852). Indicating that people rated the familiar people higher on the closeness scale. For learning value, the results for both familiar and novel was close together and the familiar options actually scored higher. The mean for familiar people was 4.207(sd.=0.927) and for novel people was 3.962(sd.=0.913).

This indicates that people thought familiar options would have a higher learning value compared to novel options. Regarding the motivation of the participants most participants indicated that the person they could choose was the most important aspect and their relationship with that person.

7.1 Effect of Age on Choice.

To examine the effect of age on choice we will first be looking how choices differ by age groups, separately for each of the scenarios, using the chi-square test. As stated in the first hypothesis we expect that in the open-ended scenarios older adults will go more for the familiar option compared to younger adults. According to the second hypothesis we expect that in the time-limited scenario both the older and younger adults will choose the familiar options more often than the novel options. When looking at the first scenario1,which was person open-ended, 68% of the young adults picked the familiar choice most often compared with 75% of the older adults, as can be seen in figure 8. The chi-square value was not

significant (p>0.05, 𝜒2=0.419) indicating that the age did not have a significant association with the choices made in the first scenario. In the second scenario, person time-limited, we see that 91% of the young adults went for the familiar choice most often compared with 97%

of the older adults, which can be seen in figure 9. Because the expected frequency

assumption has been violated (2 cells have frequencies below 5) Fisher’s Exact test is used.

Age did not have a significant association or influence on the choice which likely results from the fact that the majority of the choices for both age groups went for the familiar choice.

When looking at the third scenario, additional info time-limited shown in figure 10, 52% of the young adults chose the familiar option while of the older adults 56% went for the familiar option, with an insignificant difference (>0.05, 𝜒2=0.118). This indicates that in the third

1 This scenario consisted of the first five choices.




Related subjects :