To Live or Not to Live? The Effect of Mind Perception and Judgment Strategies on Life-Sustaining Treatment Decisions for Patients in Persistent Vegetative States

A substantial body of research suggests that the concept of personhood hinges significantly on whether a being is perceived to possess a mind [11]. Individuals’ judgments about moral status and rights often rely on perceived mental capacities, suggesting that mind perception may play a crucial role in ethical considerations related to decisions at the end of life [11, 12]. Mind perception is often inferred from how individuals evaluate an individual's consciousness and intentional actions. Conscious experience entails the capacity to sense and feel, incorporating awareness of emotional and physical states such as fear, anger, desire, hunger, or pain. Agency refers to abilities associated with planning and action, including foresight, self-control, memory, social understanding, and communication. Scores on both experience and agency tend to differ, with people attributing high scores on both to themselves, low scores to the deceased, and high experience but low agency to infants [13].

Mind perception influences medical judgments, particularly in clinical contexts where judgment of a patient's mental state informs treatment choices [14]. Failing to identify a mind in another person can reduce them to a non-person status. For example, 279 laypeople were asked to estimate the levels of agency and experience in hypothetical patients varying in degrees of consciousness and to indicate their preferences regarding the maintenance or termination of life support. These participants were more likely to opt for the discontinuation of life support for patients in a persistent vegetative state, followed by those with amyotrophic lateral sclerosis and those in a minimally conscious state. The decision to maintain life support was reliably predicted by participants' perceptions of the patient’s agency, whereas ratings of experience did not significantly influence these choices [15].

In decision-making for patients in a persistent vegetative state, three judgment strategies are typically employed: substituted judgment, best-interest standards based on community norms, and judgments rooted in respect for dignity and individuality [16]. Each strategy reflects different ethical priorities and varying interpretations of patient autonomy and dignity.

Substituted judgment corresponds to situations where surrogates make treatment decisions based on an incapacitated patient’s known desiderata [17]. This strategy dominates, despite criticism of its unreliability since patient preferences often change, and the fact that surrogate predictions are only 68% accurate. Its lasting prevalence can be ascribed to strong societal and legal support for autonomy. Although this strategy can lessen family conflicts by seemingly aligning decisions with presumed patient preferences, surrogates frequently misinterpret patients’ wishes, which raises questions as to its effectiveness in resolving ethical dilemmas [18‐20].

The best-interest standards derived from community norms [21] approach is aligned with public discourse on end-of-life care that aim to limit surrogates' discretion and prevent inappropriate treatment requests while establishing uniform care pathways. This strategy acknowledges the difficulties involved in predicting individual patient wishes and points to the importance of societal consensus on ethical treatment standards. However, achieving this type of consensus is complexified by the need for individualized care [17, 22].

The third judgment strategy based on respect for the patient’s life narrative emphasizes honoring patients’ dignity and individuality beyond mere autonomy. The life narrative reasoning model utilizes the patient’s life story and previous choices to guide surrogates in making judgments consistent with the patient's values and experiences. It aims to respect individuals by providing care that resonates on a personal level, despite the loss of their decision-making capacity [23]. However, accurately representing patient preferences without direct communication can be extremely difficult [24‐26]. In particular, the effects of different judgment strategies on decisions concerning life-sustaining treatment have not been examined empirically. This was one of the primary goals of the current set of studies.

Three experiments were conducted to provide a better understanding of the factors that influence decisions related to life-sustaining treatment for patients in a persistent vegetative state. The participants were randomly assigned to read one of three vignettes, each portraying a different level of perceived severity in the patient’s condition. After assessing the patient on the Mind Perception Scale, they were then asked to choose whether to increase, maintain, or withdraw life-sustaining treatment. To examine the role of ethical frameworks, participants were randomly assigned to one of three judgment strategies: substituted judgment, community-based norms, or life-narrative reasoning.

Study 1 tested the effects of the participants’ evaluation of a patient’s perceived severity and mind perception when asked to implement the substituted judgment strategy to decide on life-sustaining treatment.

An a-priori power analysis to estimate the required sample size (using G*Power 3.1) [27] with α = 0.05 and power = 0.80 indicated that the projected sample size needed to detect a moderate effect size (f = 0.15) was approximately N = 103 for linear multiple regression, fixed model, R2 increase [28]. In total, 281 participants were recruited through Amazon's MTurk in exchange for a small monetary reward. This sample size was adequate for the objectives of this study. All participants were included in the analysis. The participants (34% males and 66% females) ranged in age from 19 to 74 (M = 36.12; SD = 12.11).

The vignette describing a persistent vegetative state at perceived severity level 1 was as follows:

After a car accident, the patient was admitted to the hospital in an apparent coma-like state. The patient’s spinal cord was not affected during the accident, but the patient remains bedridden and immobile. The patient is unable to breathe or eat on their own and has been placed on a ventilator with a feeding tube. A series of specialized visual and motor assessments show that the patient’s basic reflexes are intact. Further tests suggest that the patient can feel and hear loved ones nearby. EEG and PET scans were performed after the patient had been comatose for twelve months.

Persistent vegetative state perceived severity level 2 provided the same information as Level 1, but ended as follows:

…. A series of specialized visual and motor assessments show that the patient’s basic reflexes are intact. Further tests suggest that the patient cannot feel or hear loved ones nearby.

Persistent vegetative state perceived severity level 3 was identical but ended:

…. A series of specialized visual and motor assessments show that the patient’s basic reflexes are not intact. Further tests suggest that the patient cannot feel or hear loved ones nearby.

Next, the participants completed the 18 items of the Dimensions of Mind-Perception scale [13] on a Likert-type scale ranging from 1 to 5 (1 = not at all, 2 = unlikely, 3 = possibly, 4 = probably, 5 = definitely). This scale is composed of 7 items addressing Agency (with a reliability coefficient of 0.94 in this study) and 11 items focusing on Experience (with a reliability coefficient of 0.97). The assessment was conducted from the perspective of what the patient could do, as indicated by statements such as "The patient you read about is capable of…". An example of an item measuring Agency is "making plans and working towards a goal". An example of an item measuring Experience is "experiencing physical or emotional pain."

The participants then decided whether to increase, maintain, or withdraw life-sustaining treatment, based on the substituted judgment strategy. Specifically, the participants were asked to make a decision about the patient’s treatment as though they were in their shoes. They were asked to rate each outcome (increase treatment, maintain treatment, withdraw treatment) on a 6-point Likert scale.

To test whether the perceived severity of the condition described in the vignette affected the participants' decision to increase, maintain or withdraw treatment, a multivariate analysis of variance (MANOVA) was conducted and revealed a significant multivariate effect for perceived severity (F(6, 552) = 5.34, p < 0.001, Wilk’s l = 0.89 ηp² = 0.06) on the set of dependent measures. Significant between-subjects effects were found for the choice to increase treatment (F(2, 278) = 13.94, p < 0.001, ηp² = 0.09), maintain treatment (F(2, 278) = 10.89, p < 0.001, ηp² = 0.07) and for withdraw treatment (F(2, 278) = 10.30, p < 0.001, ηp² = 0.07). Bonferroni post-hoc tests showed that on average, the participants were significantly less likely to support increasing treatment as condition severity increased, with all p’s ranging from < 0.001 to 0.035). Table 1 presents the means and standard deviations of the ranking participants ascribed to each decision in the three perceived severity levels.

Bonferroni post-hoc tests showed a decrease in willingness to maintain treatment as the perceived severity of the condition increased. Participants who read the least perceived severe condition vignette expressed the highest willingness to increase treatment compared to the high severity group who read vignette 3. The mean scores only differed significantly between the most perceived severe and the least perceived severe conditions (p < 0.001). The mean scores were only marginally significant for the high perceived severity vs. the moderate perceived severity case (p < 0.052) and for the moderate perceived severity case vs. the low perceived severity case (p = 0.068) (see Table 1).

As regarding the willingness to withdraw treatment, Bonferroni post-hoc tests showed that participants who read the least perceived severe condition vignette showed the least willingness to withdraw treatment than participants who read the more severe scenarios (pmoderate = 0.018; pmost severe < 0.001). The mean differences for withdrawing treatment were not significant between the high perceived severity and moderate perceived severity conditions (p = 0.241) (see Table 1).

To examine the effects of Agency and Experience on the participants’ decisions, a correlational analysis followed by an analysis of the moderating roles of agency and experience on the association between the perceived severity of the condition and the likelihood of each decision. The means, standard deviations and the Pearson inter-correlations for the main variables are presented in Table 2.

As shown in Table 2, the decisions to maintain or increase treatment were highly positively correlated and both were negatively correlated with the decision to withdraw treatment. Both Agency and Experience were positively correlated with the decisions to maintain or increase treatment and negatively to withdraw treatment. In addition, the greater the perceived severity of the condition, the higher the reported levels of Agency and Experience. The greater the perceived severity of the condition, the less likely participants were to decide to maintain or increase treatment, and the more likely they were to decide to withdraw it. Age was negatively correlated with the decision to maintain or increase treatment and negatively correlated with Agency.

To further examine the moderating role of mind perception on the association between perceived severity of the condition and the likelihood of a given decision, PROCESS macro Model 1 [29] was used. The results revealed that in addition to the associations in Table 2, the only significant interaction was between Agency and Experience in predicting the likelihood to withdraw treatment, but not the decision to increase or maintain treatment. Table 3 presents the results.

Table 3 shows the significant interaction between agency and experience in predicting the decision to withdraw treatment. Simple slope analyses revealed that a significant negative correlation was only found between agency and the likelihood to withdraw treatment when experience was low. In other words, the mind perceptions of both Experience and Agency needed to be low to predict a greater likelihood to make the decision to withdraw treatment.

Study 2 was designed to examine the impact of perceived severity of the condition and mind perception when participants were asked to implement the community-based norms strategy.

We employed the same a-priori power analysis procedure as in Study 1 to determine the required sample size (using G*Power 3.1) [27]. In total, 271 Mturk participants were recruited. This sample size was adequate for the objectives of this study. All participants were included in the analyses. The participants (33.6% males and 66.4% females) ranged in age from 18 to 73 (M = 37.42; SD = 12.81).

The procedure was identical to Study 1, except for the change in judgment strategy. Participants evaluated the patient's mind perception on the Mind Perception scale [13]. Agency had a reliability coefficient of 0.95 in this study and Experience had a reliability coefficient of 0.98. The participants then made the decision whether to increase, maintain, or withdraw life-sustaining treatment based on the community norms-based strategy on a 6-point Likert scale.

To test whether perceived severity affected the participants’ decision to increase, maintain or withdraw treatment, a multivariate analysis of variance (MANOVA) was conducted and revealed a significant multivariate effect for perceived severity (F(6, 532) = 14.16, p < 0.001, Wilk’s l = 0.74 ηp² = 0.14) on the set of dependent measures. Between-subjects tests revealed significant effects for the decision to increase treatment (F(2, 268) = 18.21, p < 0.001, ηp² = 0.12), maintain treatment (F(2, 268) = 23.71, p < 0.001, ηp² = 0.15) and withdraw treatment (F(2, 268) = 37.83, p < 0.001, ηp² = 0.22). Bonferroni post-hoc tests showed that participants were significantly less likely to support increasing treatment as condition severity increased. Specifically for the low perceived severity vignette, their intention to increase treatment was higher than when the patient’s perceived severity condition was presented as moderate. The lowest intention to increase treatment was found for participants evaluating the severe condition vignette (all p’s ranged from < 0.001 to 0.024; see Table 1). Bonferroni post-hoc tests revealed that participants were less likely to maintain treatment when the condition was perceived as the most severe. The participants’ decision to increase treatment was higher when the patient’s condition was presented as moderate and the lowest when the patient’s condition was perceived as severe. By contrast to the substituted judgement strategy, in the community norms strategy, the three levels of severity produced significantly different decision outcomes (all p’s < 0.001) except for a non-significant difference between the means for moderate perceived severity and low perceived severity, where the decision to maintain treatment was similar and higher than in the high perceived severity vignette (see Table 1). Again, unlike in Study 1, Bonferroni post-hoc tests revealed that in the community norms strategy, the decision to withdraw treatment varied significantly across all severity conditions (all p’s < 0.001). Specifically, the greater the perceived severity, the more the participants were likely to decide to withdraw the treatment (see Table 1).

To examine the effects of Agency and Experience on the participants’ decisions, a correlational analysis followed by an analysis of the moderating roles of Agency and Experience on the association between the perceived severity of the condition and the likelihood of each decision were conducted. The means, standard deviations and Pearson inter-correlations between the main variables are presented in Table 4.

As shown in Table 4, the decisions to maintain or increase treatment were highly positively correlated and both were negatively correlated with the decision to withdraw treatment. Both Agency and Experience were positively correlated with the decisions to maintain or increase treatment and negatively to withdraw treatment. In addition, the greater the perceived severity of the condition, the higher the reported levels of Agency and Experience. The greater the perceived severity of the condition, the less likely the participants were to decide to maintain or increase treatment, and the more likely they were to decide to withdraw it. Age was negatively correlated with the decision to increase treatment and with Agency. Thus, the results were similar to those found in Study 1.

To further examine the moderating role of mind perception on the association between perceived severity of the condition and the likelihood of each decision, PROCESS macro Model 3 [29] was used. Table 5 presents the results.

As shown in Table 5, there was a significant interaction between Agency, Experience and the patient’s perceived severity in predicting the withdrawal decision. Simple slope analyses revealed that when both agency and experience were low or high, and when agency was moderate and experience was low, the greater the perceived severity, the more the participants were likely to recommend withdrawing treatment (Table 5).

Study 3 was designed to examine the effect of perceived severity and mind perception when participants were asked to implement the life narrative strategy in treatment decision-making.

We used the same a-priori power analysis to estimate the required sample size as in Studies 1 and 2 (using G*Power 3.1) [27]. In total, 263 Mturk participants were recruited. This sample size was adequate for the objectives of this study. All participants were included in the analyses. The participants (32.4% males and 67.6% females) ranged in age from 18 to 74 (M = 34.89; SD = 11.69).

The procedure was identical to Study 1, except for the change in judgment strategy. Participants evaluated the patient's mind perception [13], Agency (with a reliability coefficient of 0.92 in this study) and Experience (with a reliability coefficient of 0.94), and then made decisions whether to increase, maintain, or withdraw life-sustaining treatment, based on the patient’s life narrative reasoning strategy on a 6-point Likert scale.

To test whether the perceived severity of the condition affected the participants’ decision to increase, maintain or withdraw treatment, a multivariate analysis of variance (MANOVA) was conducted and revealed a significant multivariate effect for perceived severity (F(6, 516) = 9.82, p < 0.001, Wilk’s l = 0.81 ηp² = 0.10) on the set of dependent measures. Between-subject tests revealed significant effects for the decision to increase treatment (F(2, 260) = 16.51, p < 0.001, ηp² = 0.11) maintain treatment (F(2, 260) = 24.60, p < 0.001, ηp² = 0.16) and withdraw treatment (F(2, 260) = 21.66, p < 0.001, ηp² = 0.14). Bonferroni post-hoc tests revealed that in contrast to the other two strategies, when the strategy was respect for the patient’s life narrative reasoning, there were no significant differences between the severe and moderate cases in the decision to increase treatment (p = 0.153); namely, both conditions resulted in lower support compared to the least severe case (all p’s < 0.001) (see Table 1). Moreover, unlike the previous strategies, Bonferroni post-hoc tests indicated no significant difference between the perceived most severe and perceived moderate conditions for maintaining treatment (p = 0.112), but in both cases the participants were less willing to maintain treatment in comparison to the low perceived severity vignette (all p’s < 0.001) (see Table 1).

Bonferroni post-hoc tests revealed that like the community-based norms strategy tested in Study 2, but unlike the substitute judgement strategy in Study 1, all the severity conditions were significantly different from one another (all p’s ranged from < 0.001 to.007). Specifically, the greater the perceived severity, the more the participants decided to withdraw treatment (see Table 1).

To examine the effects of Agency and Experience on the participants’ decisions, a correlational analysis followed by an analysis of the moderating roles of agency and experience on the association between the perceived severity of the condition and the likelihood of each decision were conducted. The means, standard deviations and the Pearson inter-correlations between the main variables are presented in Table 6.

As shown in Table 6, the decisions to maintain or increase treatment were highly positively correlated, and both were negatively correlated with the decision to withdraw treatment. Both Agency and Experience were positively correlated with the decisions to maintain or increase treatment and negatively with the decision to withdraw treatment. In addition, the greater the perceived severity of the condition, the higher the reported levels of Agency and Experience. The greater the perceived severity of the condition, the less likely participants were to decide to maintain or increase treatment, and the more likely they were to decide to withdraw it. These results were similar to those found in Studies 1 and 2 apart from age.

To further examine the moderating role of mind perception on the association between perceived severity and the likelihood of each decision, PROCESS macro Model 1 [29] was implemented. Table 7 presents the results.

As shown in Table 7, there was a significant interaction between Agency and Experience in predicting both the decision to increase and to maintain treatment. Simple slope analyses revealed that in the case of the decision to increase treatment, the association between Agency and the decision to increase treatment was positively significant for all Experience levels but was higher with lower Experience. In terms of the decision to maintain treatment, there was only a significant positive association between Agency and the decision to maintain treatment when Experience was low. No other interactions were found.

Because this study used hypothetical scenarios that were submitted to lay participants, the findings may not fully translate to real-world clinical decision-making contexts [5]. The predominantly American sample limits the applicability of the findings to other cultural contexts where norms and ethical values may differ substantially. It is also worth noting that the absence of significant differences between the ethical strategies suggests that each presents practical challenges. Current tools for assessing or simulating patient preferences may not adequately support reliable decision-making for incapacitated individuals. While the reliance on intuitive processing offers one explanation, another possibility is that the ethical strategies were not sufficiently internalized or salient to guide participants' decisions effectively. Although each strategy was introduced with a clear and concise description, it is plausible that in the context of a brief experimental vignette, participants reverted to more emotionally accessible cues such as perceived responsiveness over more abstract reasoning tools. This may reflect a general limitation in how well such strategies capture hypothetical judgments under uncertainty. In addition, the emotional intensity and moral gravity of the scenarios may have overwhelmed participants’ capacity or motivation to apply formal decision rules, particularly without extended reflection or training. Taken together, these findings suggest both a default reliance on perceived severity and potential limitations in the operationalization and uptake of structured ethical strategies in experimental settings.

Future research could examine whether more immersive simulations, repeated exposure, or scaffolded reflection could enhance participants' ability to engage with ethical reasoning frameworks in complex moral decisions. Beyond the empirical findings, these results have broader ethical and policy implications. The limited effectiveness of the substituted judgment strategy calls into question the extent to which laypeople can reliably reconstruct patient preferences although this assumption underpins legal standards in many jurisdictions. The intuitive dominance of perceived responsiveness over value-based reasoning suggests that frameworks emphasizing dignity, narrative identity, or community norms may face practical challenges in implementation without greater public understanding or facilitation.

Perhaps the most concerning factor is the potential reliance on observable behavior as a proxy for consciousness, particularly considering the known risks of misdiagnosis in vegetative and minimally conscious states [33]. If lay decision-makers treat reflexive or emotional responses as definitive signs of mind, this could lead to ethically problematic decisions, especially in ambiguous or borderline cases. Future models must therefore trade off intuitive moral sensitivity with safeguards that account for diagnostic uncertainty and the complexity of human consciousness.

Although the present study did not investigate artificial intelligence (AI), the findings may contribute to broader discussions about future methods for reconstructing patient preferences. One such approach is the Patient Preference Predictor (PPP), which aims to forecast the choices of incapacitated patients using demographic data from similar individuals. Nonetheless, this approach has drawn criticism for overlooking the particular values and priorities of individual patients [34].

To address these concerns, a more advanced theoretical model has been proposed: the Personalized Patient Preference Predictor (P4) [34]. Unlike the PPP, this model seeks to reconstruct patient preferences based on individualized data such as previous medical decisions, legal documents, treatment history, and even digital communication. Its goal is not only to predict what the patient would choose, but to capture the reasoning, values, and priorities that informed those choices. Our findings suggest that laypeople rely on observable cues as proxies for mind, especially under conditions of diagnostic uncertainty. This raises the question of whether AI tools like the Personalized Patient Preference Predictor (P4) can mitigate such biases by offering data-driven insights into patient preferences. However, the ethical legitimacy of these models depends not only on predictive accuracy, but on their capacity to capture the patient’s underlying values and contextual reasoning—a challenge that mirrors the difficulty our participants faced in applying structured ethical reasoning. As such, integrating AI into surrogate decision-making processes must address both informational and moral dimensions to be ethically defensible. In doing so, it aspires to uphold autonomy even in the absence of direct communication. However, any potential application of AI in ethical decision-making must be approached with caution. Some models function as "black boxes," offering little transparency about their internal logic, which can undermine legitimacy and public trust [35]. Critics have argued that even highly accurate predictions may fall short of true autonomy if the decision lacks an understandable or value-based justification [36].