← Return to search results
Back to Prindle Institute

grief

Posted on

Resurrection Through Chatbot?

cartoon image of an occult seance

There is nothing that causes more grief than the death of a loved one; it can inflict an open wound that never fully heals, even if we can temporarily forget that it’s there. We are social beings and our identities aren’t contained within our own human-shaped space. Who we are is a matter of the roles we take on, the people we care for, and the relationships that allow us to practice and feel love. The people we love are part of who we are and when one of them dies, it can feel like part of us dies as well. For many of us, the idea that we will never interact with our loved one again is unbearable.

Some entrepreneurs see any desire as an opportunity, even the existential impulses and longings that come along with death. In response to the need to have loved ones back in our lives, tech companies have found a new use for their deepfake technology. Typically used to simulate the behavior of celebrities and politicians, some startups have recognized the potential in programming deepfake chat-bots to behave like dead loved ones. The companies that create these bots harvest data from the deceased person’s social media accounts. Artificial intelligence is then used to predict what the person in question would say in a wide range of circumstances. A bereaved friend or family member can then chat with the resulting intelligence and, if things go well, it will be indistinguishable from the person who passed away.

Some people are concerned that this is just another way for corporations to exploit grieving people. Producers of the chatbots aren’t interested in the well-being of their clients, they’re only concerned with making money. It may be the case that this is an inherently manipulative practice, and in the worst of ways. How could it possibly be acceptable to profit from people experiencing the lowest points in their lives?

That said, the death industry is thriving, even without the addition of chatbots. Companies sell survivors of the deceased burial plots, coffins, flowers, cosmetic services, and all sorts of other products. Customers can decide for themselves which goods and services they’d like to pay for. The same is true with a chatbot. No one is forced to strike up a conversation with a simulated loved one, they have a chance to do so only if they have decided for themselves that it is a good idea for them.

In addition to the set of objections related to coercion, there are objections concerning the autonomy of the people being simulated. If it’s possible to harm the dead, then in some cases that may be what’s going on here. We don’t know what the chatbot is going to say, and it may be difficult for the person interacting with the bot to maintain the distinction between the bot and the real person they’ve lost. The bot may take on commitments or express values that the living person never had. The same principle is at play when it comes to using artificial intelligence to create versions of actors to play roles. The real person may never have consented to say or do the things that the manufactured version of them says or does. Presumably, the deceased person, while living, had a set of desires related to their legacy and the ways in which they wanted other people to think of them. We can’t control what’s in the heads of others, but perhaps our memories should not be tarnished nor our posthumous desires frustrated by people looking to resurrect our psychologies for some quick cash.

In response, some might argue that dead people can’t be harmed. As Epicurus said, “When we exist, death is not; and when death exists, we are not. All sensation and consciousness ends with death and therefore in death there is neither pleasure nor pain.” There may be some living people who are disturbed by what the bot is doing, but that harm doesn’t befall the dead person — the dead person no longer exists. It’s important to respect autonomy, but such respect is only possible for people who are capable of exercising it, and dead people can’t.

Another criticism of the use of chat-bots is that it makes it more difficult for people to arrive at some form of closure. Instead, they are prolonging the experience of having the deceased with them indefinitely. Feeling grief in a healthy way involves the recognition that the loved one in question is really gone.

In response, some might argue that everyone feels grief differently and that there is no single healthy way to experience it. For some people, it might help to use a chat-bot to say goodbye, to express love to a realistic copy of their loved one, or to unburden themselves by sharing some other sentiment that they always needed to let out but never got the chance.

Other worries about chatbot technology are not unique to bots that simulate the responses of people who have passed on. Instead, the concern is about the role that technology, and artificial intelligence in particular, should be playing in human lives. Some people, will, no doubt, opt to continue to engage in a relationship with the chat-bot. This motivates the question: can we flourish as human beings if we trade in our interpersonal relationships with other sentient beings for relationships with realistic, but nevertheless non-sentient artificial intelligence? Human beings help one another achieve the virtues that come along with friendship, the parent-child relationship, mentorship, and romantic love (to name just a few). It may be the case that developing interpersonal virtues involves responding to the autonomy and vulnerability of creatures with thoughts and feelings who can share in the familiar sentiments that make it beautiful to be alive.

Care ethicists offer the insight that when we enter into relationships, we take on role-based obligations that require care. Care can only take place when the parties to the relationship are capable of caring. In recent years we have experimented with robotic health care providers, robotic sex workers, and robotic priests. Critics of this kind of technological encroachment wonder whether such functions ought to be replaced by uncaring robots. Living a human life requires give and take, expressing and responding to need. This is a dynamic that is not fully present when these roles are filled by robots.

Some may respond that we have yet to imagine the range of possibilities that relationships with artificial intelligence may provide. In an ideal world, everyone has loving, caring companions and people help one another live healthy, flourishing lives. In the world in which we live, however, some people are desperately lonely. Such people benefit from affection behavior, even if the affection is not coming from a sentient creature. For such people, it would be better to have lengthy conversations with a realistic chat-bot than to have no conversations at all.

What’s more, our response to affection between human beings and artificial intelligence may say more about our biases against the unfamiliar than it does against the permissibility of these kinds of interactions. Our experiences with the world up to this point have motivated reflection on the kinds of experiences that are virtuous, valuable, and meaningful. Doing so has necessitated a rejection of certain myopic ways of viewing the boundaries of meaningful experience. We may be at the start of a riveting new chapter on the forms of possible engagement between carbon and silicon. For all we know, these interactions may be great additions to the narrative.

Posted on

The Knowledge Norms of Emotions

simple single-line drawing of person curled up in bed

This post begins with a sad backstory. A little while back my wife and I had a miscarriage of our first child. There was a lot that was terrible in the experience, but in this post I want to address a certain oddity that I noticed about grief.

Due to a range of bureaucratic complications, it took about a week from when we first suspected a miscarriage to when we had final confirmation. During that week, our confidence that we had miscarried grew, but throughout the period it remained a likelihood rather than a certainty.

What surprised me, during that week, was that the uncertainty made it difficult to grieve. Even when I was pretty sure we had lost the child, it felt ‘out of whack’ to grieve the loss, since there was a chance the child was still alive. It was a terrible week, and I was extremely sad, but it felt out of joint to grieve for the child while recognizing the chance that all might be well. There was no obstacle to feeling anxious, there was nothing out of joint about feeling trepidation, but outright grief felt strange. And it continued to feel strange until we received confirmation of the loss.

This, eventually, got me wondering: is grief characterized by a knowledge norm? In philosophy, a knowledge norm is a normative rule which says that knowledge of something is required for an action or mental state to be appropriate. For example, there seems to be a knowledge norm on assertion: you should only tell someone something if you know that thing is true. This explains, for instance, why if I say “it will rain tomorrow” it is appropriate for you to ask “how do you know?” Or why saying “I don’t know” is an appropriate response if someone asks you a question. (For a thorough defense of a knowledge norm of assertion see Timothy Williamson’s “Knowing and Asserting.”)

Many philosophers also argue that there is a knowledge norm of belief: you should only believe X if you know X is true. Thus, Williamson argues in his book Knowledge and its Limits: 

“Knowledge sets the standard of appropriateness for belief. That does not imply that all cases of knowing are paradigmatic cases of believing, for one might know p while in a sense treating p as if one did not know p—that is, while treating p in ways untypical of those in which subjects treat what they know. Nevertheless, as a crude generalization, the further one is from knowing p, the less appropriate it is to believe p. Knowing is in that sense the best kind of believing. Mere believing is a kind of botched knowing. In short, belief aims at knowledge (not just truth).”

There also seems to be a knowledge norm of certain actions. For instance, it seems like you should only punish someone if you know they are guilty, and only chastise someone if you know they did wrong. Some philosophers have gone even further and suggested that there is a general knowledge norm on all action: you should only treat X as a reason for action if you know X to be true.

My own experience with grief seems to suggest that there might also be a knowledge norm on various emotions; but as far as I know that topic has not yet been seriously investigated by philosophers.

My experience of the miscarriage suggested there was a knowledge norm to grief because the problem was that it felt wrong to grieve our child’s death as long as I recognized that the child might still be alive. This is parallel to how I can’t know the child had died as long as I recognized that the child might still be alive. In some sense, what is characteristic of knowledge is the elimination of all relevant alternatives. As long as those relevant alternatives remained, we did not know, nor did it feel quite right to grieve.

Here is another reason for thinking that grief is characterized by a knowledge norm: it is hard to fit probabilities with the emotion of grief. It would be weird to think that as I grow more certain, my grief grows proportionally. I do not grieve a small amount at a 5% chance that my spouse has died, nor would my grief double as my confidence grows to 10%. I grieve less for less bad things, not for lower probabilities of equally bad things. But it would be equally weird to think that there is some probabilistic threshold at which grief suddenly becomes appropriate. It is not as though when I go from 94% confident my child died to 96% confident my child died that suddenly grief goes from inappropriate to appropriate.

But if grief neither scales with probability, nor requires a certain probabilistic threshold, then it seems like grief is responsive to a standard other than probabilistic credence, and the natural alternative is that it is responsive to knowledge.

Other emotions also seem to be knowledge normed in this way. It is hard to feel grateful because you think it is likely that someone brought you a present. Normally gratitude is a response to the knowledge that someone did something for you. Jonathan Adler makes a point along these lines about resentment: “Mild resentment is never resentment caused by what one judges to be a serious offense directed toward oneself tempered by one’s degree of uncertainty in that judgment.”

Now, some other emotions at first blush seem different. I can be worried about something without knowing that thing will occur. Similarly, I can be hopeful of something without knowledge it will occur. Yet, even here, it seems that there might be some knowledge norm at play. For instance, it seems weird to be worried about or hope for something you know is impossible. Thus, it might be that you must know that something is possible before you can worry about it or hope for it.

If this is right, does it suggest a general pattern? I think it does. Emotions have appropriateness conditions. Resentment is an appropriate response to being wronged. Gratitude is an appropriate response to being given a gift. Hope is an appropriate response to the possibility of certain goods, as worry is an appropriate response to the possibility of certain bads. In each of these cases, what is required to rightly feel the emotion is knowledge.

That, then, is why grieving felt strange. I didn’t yet know if my grief was appropriate since I lacked knowledge of the tragedy to which my grief was a response.

Posted on

A Problem with Emotions

abstract acrylic painting of divided canvas

There is a certain challenge to the adequacy of our emotional reactions — especially those reactions, like grief and joy, which feel ‘called for’ at certain times. Suppose a family has a child who falls grievously ill. After many sleepless nights, the child stabilizes and eventually recovers. There are appropriate emotional responses to this sequence; the parents will, and should, feel relieved and joyed at the child’s recovery. Now suppose another family has a child who similarly falls grievously ill. Except this child does not recover and eventually dies. Again, there are appropriate emotional responses. The parents will, and should, feel grieved and heartbroken at the child’s death.

So far, there is no challenge. But now suppose that instead of two different families, it was one family with two children — one recovers, one dies. Here, what are the parents supposed to feel? There are a couple of options.

Perhaps they should feel a sort of moderated grief. After all, something wonderful has happened (a child has recovered) and something terrible has happened (a child has died). Do they partially cancel out (but maybe weighted in the direction of grief since ‘bad is stronger than good’)? The problem with this answer is that the grief is a response to the tragedy of the child’s death. And that child’s death is no less a tragedy just because the other child survived. Moderation would be appropriate if something happened to moderate the tragedy of the child’s death — such as the child being spared death and instead placed within an enchanted sleep — but it does not seem like the appropriate response to some other good thing occurring.

Perhaps then, you just need to feel either emotion. Both grief and joy are appropriate — so long as you feel one, then you are feeling well. But this won’t do either. There is something wrong with the parent who feels nothing for the recovery of their child, just as there is something wrong with the parent who feels nothing for the child’s death.

In fact, the only response that seems appropriate to the situation is to feel both grief and joy. You ought to be grieved at the one child’s death and joyed at the other child’s recovery.

But here is the issue. It doesn’t seem possible to fully feel both at once. Feelings, unlike some other mental states, compete with each other. When I feel happy about one thing, it pushes sadness about other things to the periphery. This is unlike, say, beliefs. The parents can fully believe that one child recovered while, at the same moment, fully believing that the other child died. This is because beliefs do not require active attention. Moments ago, you believed all sorts of things about your former elementary school, but I expect until you read this sentence you were not actively attending to any of those beliefs.

Emotions, however, do require attention. If I can become fully absorbed in my work, then for a time my grief will retreat. (Of course, one of the frustrating things about grief is the way that it maintains a ‘grip’ on your attention — forcing your thoughts to circle back and return again, and again, to the tragedy.)

So, to fully feel the grief at the one child’s death, and to fully feel the joy at the other child’s recovery, would require me to keep my full attention on both at the same time. But we can’t do that with attention, attention is a limited resource. It can only be fully engaged in one direction.

The best we can do, then, is a sort of ping-ponging back and forth between grief and joy. Feeling complete grief when attending to the death, feeling thankful and relieved when attending to the recovery. But at no point, it seems, can my emotions be completely responsive to what is called for.

Berislav Marušić, in his essay “Do Reasons Expire”, considers a related puzzle:

“Grief is, plausibly, a response to reasons; the reason for my grief was my mother’s death; her death does not change over time; but it is not wrong for me to grieve less over time. Yet how could the diminution of grief not be wrong, if my reason for grief stays the same?”

The reason the problem is similar is that there is a disconnect between the response demanded by the event (the tragedy of someone’s death) and the psychological realities of our capacity to have emotions. You just can’t indefinitely grieve, and in turn you don’t indefinitely grieve. But doesn’t it seem as if there is a sense in which you ought to?

There is a conflict, then, between the psychological realities that constraint our emotions, and the appropriateness conditions surrounding what emotions we ‘ought’ to feel.

This is an important conflict to think about. One reason it’s important to be aware of this conflict is because it helps recognize exactly why we need to be so skeptical of grounding our moral decisions simply on emotions like anger or grief. Since we can only feel some emotions to an extent, our emotional responses, at a given time, are usually not responsive to the full range of relevant considerations. You can feel outrage about an injustice, or hopeful at political progress that has been made, but you can’t feel both at the same time to the appropriate extent. But given that psychological reality, that means that basing policy recommendations on emotions of rage or optimistic hope is likely to be morally dangerous.

This does not mean that emotions should play no role in our moral decision-making. Emotions are important. Instead, what this means is that we need to be extremely cautious when acting on our emotional reactions. We should always bear in mind that emotions are likely to not be reflective of the full range of complexities in any given case.

Posted on

Grief and Saint Augustine (and WandaVision)

image of failed tv signal noise

WARNING: The following article contains spoilers for all nine episodes of WandaVision on Disney+.

In 2019, Martin Scorsese ruffled fan-feathers when he explained why he doesn’t watch the movies in the Marvel Cinematic Universe: “I tried, you know? But that’s not cinema….It isn’t the cinema of human beings trying to convey emotional, psychological experiences to another human being.” This particular sentence might sound odd to those watching the latest entry in the MCU: the 9-part limited series WandaVision on Disney+, which aims to explore experiences of intense grief and loss (even as it offers up yet another batch of costumed superheroes tossing about punches and witty one-liners).

First introduced in 2015’s Avengers: Age of Ultron, Wanda Maximoff is a super-powered magic user who (along with her brother) betrayed her villainous compatriots to assist the Avengers in saving the world. In the same film, several magical items came together inside a high-tech regeneration chamber, creating Vision, an android who could fly, fire beams of cosmic energy, and alter the density of his molecules to phase through solid matter at will. Over the course of several films, the two characters grew close and fell in love, but their relationship ended tragically when Vision sacrificed himself (at Wanda’s hand) to prevent the death of half the universe at the end of Avengers: Infinity War.

As fans of the MCU know, the story is a little more complicated than that (for example: Vision’s sacrifice turned out to be in vain, though his surviving teammates later managed to undo most of the damage in Avengers: Endgame). But the events of WandaVision begin with Wanda racked with guilt over killing Vision and mourning her many losses: her parents died when she was ten, her brother died in the climax of Age of Ultron, her powers precipitated the catastrophe that sparked the events of Captain America: Civil War, and Vision (because of some time-traveling) actually died twice at the end of Infinity War. In response to all of this, Wanda’s reality-altering powers accidentally engulf the town of Westview, New Jersey, warping it into a pastiche of various television sitcoms that Wanda enjoyed with her family as a child. Within this waking dream, Wanda is not only reunited with a reconstituted (though memory-less) Vision, but the now-happily-married couple also welcomes the birth of twin sons.

(Explaining superhero stories always sounds a bit odd, doesn’t it?)

The point is that, in various ways, WandaVision seeks to explore the painful consequences of loss and other traumas. Rather than shying away from the psychological damage done to survivors of death and terror, the show centers the experience of several characters grappling with the pain of prematurely saying goodbye to those they love. Wanda’s grief over losing Vision is mirrored in the storyline of Monica Rambeau, first introduced in 2019’s Captain Marvel and now working as an intelligence agent. Midway through the series, the audience learns that Monica was one of the people who Wanda failed to save by killing Vision in Infinity War (but who was also resurrected five years later by the Avengers in Endgame). During the interim, Monica’s mother died of cancer — something Monica learns mere minutes after returning to life and mere days before encountering Wanda in Westview.

In the penultimate episode, an antagonist leads Wanda through several of her own memories, forcing her to confront many of the most traumatic moments in her life (including the death of her parents). During these flashbacks, a scene from the early days of Wanda and Vision’s relationship took the internet by storm: while comforting Wanda after the death of her brother, Vision encourages her that even within the waves of grief buffeting her in her loss, there must still be something good: “It can’t all be sorrow, can it? I’ve always been alone, so I don’t feel the lack — it’s all I’ve ever known. I’ve never experienced loss because I’ve never had a loved one to lose. But what is grief if not love persevering?”

As you might imagine, philosophers have some other answers to this question.

Sometimes, philosophers have discussed grief as a hindrance or distraction from the “proper” objects of our attention. Consider Seneca, the Roman Stoic, who advised the daughter of a dead man to “do battle with your grief” by considering the most logical approach to find peace after her loss:

“…if the dead cannot be brought back to life, however much we may beat our breasts, if destiny remains fixed and immovable forever, not to be changed by any sorrow, however great, and death does not loose his hold of anything that he once has taken away, then let our futile grief be brought to an end.”

Often depicted (not undeservedly, at times) as unfeeling or cold, the Stoics sought to control their emotions (and all other impulses) so as to live a life governed entirely by reason. This did not mean that the Stoics considered grief (or other emotions) inherently bad, but rather that they saw how emotional dysregulation of any kind could upset the careful balance of human psychology. Certainly, at its worst, grief can threaten to overwhelm us — just as Wanda Maximoff’s story depicts.

On the other hand, philosophers have sometimes described grief or sorrow as simply constitutive of the human experience. For thinkers like Nietzsche or Schopenhauer, the painfulness of human existence meant that sorrow and loss was simply unavoidable, so the strong must confront their grief and bend it to their will. For philosophers with a more religious or existentialist bent, the reality of grief might be borne from the sinfulness of a broken Creation or from the failure of free creatures to grapple with their own mortality. Consider how the 18th century philosopher and theologian Søren Kierkegaard explained “My sorrow/grief is my baronial castle, which like an eagle’s nest is built high up on the mountain peak among the clouds. No one can take it by storm.” On these perspectives, grief is not something that can even possibly be dissolved, but rather must be harnessed and (hopefully) understood.

WandaVision’s treatment of grief is a line between these extremes: neither rejecting the emotion as inappropriate nor reveling in it as inevitable. It is a line akin to the picture found in the work of St. Augustine of Hippo, who describes in his autobiographical Confessions how the death of a loved one caused him such great distress that he nearly felt like he would die himself. Because of the love he felt for this unnamed friend (“I had felt that my soul and his soul were ‘one soul in two bodies,’” he says in IV.vi.11), Augustine was devastated by his death; regardless of death’s inevitability, “The lost life of those who dies becomes the death of those still living” (IV.ix.14).

And although Augustine (much like his Stoic forebears) infamously sought to curtail the public expressions of his grief after his conversion to Christianity (lest he suggest that the state of a departed soul was not improved by its transition to the afterlife or, even worse, pridefully demand the solace of others), Augustine never argues that grief is, in principle, sinful. In a particularly vulnerable passage, Augustine confesses how, after the death of his own mother, he found a private place and “let flow the tears which I had held back so that they ran as freely as they wished” (IX.xii.33). His love for his loved one persevered (and, in fact, drove him to an even deeper love for God).

Ultimately, WandaVision ends with Wanda realizing how her uncontrolled grief has led her to hurt the people of Westview (something a more Stoic approach to death would have avoided). Tearfully, she accepts (along with Kierkegaard) the inevitability of her pain and chooses to free the town by saying goodbye to her imaginary loved ones. But, just as Wanda’s memories and magic remain within her, so too does her love persevere; in their final moments together, the dream-Vision encourages Wanda once again: “We have said goodbye before, so it stands to reason–” at which point, Wanda sobs “…we’ll say hello again.”

Saint Augustine would indeed agree; the only real problem with Wanda’s grieving love was how she chose to express it.

In an attempt to clarify his criticism of the MCU, Martin Scorsese later published an op-ed in The New York Times where he explained how he always believed that “cinema was about revelation — aesthetic, emotional and spiritual revelation. It was about characters — the complexity of people and their contradictory and sometimes paradoxical natures, the way they can hurt one another and love one another and suddenly come face to face with themselves.” To be blunt, on such a definition, it’s hard to see how the love and pain of Wanda Maximoff fails to qualify.

And, unbeknownst to Wanda, several lingering plot threads suggest that hope does indeed remain for a genuine family reunion, but fans will have to wait for future MCU installments to see what happens next. In the meanwhile, it stands to reason that we might all benefit from reading a little more philosophy (and not just the bits about “identity metaphysics”) to help us think through our own complicated experiences of grief (and love).

Posted on

Shared Grief Does Not Always Unite

The past few weeks have been hard for those who are fervently anti-Trump. On the weekend after the election, I was playing with my baby daughter, and made a comment about how empathetic I am.

My partner, who was lying on the couch next to me, muttered sarcastically: “Why don’t you go empathize with the white working class.”

My reaction was immediate, unreflective, and dramatic: I started shouting at him. That comment was uncalled for, utterly gratuitous! I was on the same side as his! I in no way thought that white men were more deserving of empathy than others, as I took him to imply. Finally, I started using expletives, and told him to f*ck off.

Yes, I told my beloved partner, a man of color who has been grieving the electoral result and has found it hard to get out of bed since then, that he could f*ck off.

Continue reading “Shared Grief Does Not Always Unite”