Folk Psychological Attributions of Consciousness to Large Language Models

Colombatto, C., & Fleming, S. M.

(2023, November 22). PsyArXiv

https://doi.org/10.31234/osf.io/5cnrv

Abstract

Technological advances raise new puzzles and challenges for cognitive science and the study of how humans think about and interact with artificial intelligence (AI). For example, the advent of Large Language Models and their human-like linguistic abilities has raised substantial debate regarding whether or not AI could be conscious. Here we consider the question of whether AI could have subjective experiences such as feelings and sensations (“phenomenological consciousness”). While experts from many fieldshave weighed in on this issue in academic and public discourse, it remains unknown how the general population attributes phenomenology to AI. We surveyed a sample of US residents (N=300) and found that a majority of participants were willing to attribute phenomenological consciousness to LLMs. These attributions were robust, as they predicted attributions of mental states typically associated with phenomenology –but also flexible, as they were sensitive to individual differences such as usage frequency. Overall, these results show how folk intuitions about AI consciousness can diverge from expert intuitions –with important implications for the legal and ethical status of AI.

The preprint is here.

My summary:

The results of the study show that people are generally more likely to attribute consciousness to LLMs than to other non-human entities, such as animals, plants, and robots. However, the level of consciousness attributed to LLMs is still relatively low, with most participants rating them as less conscious than humans. The authors argue that these findings reflect the influence of folk psychology, which is the tendency to explain the behavior of others in terms of mental states.

The authors also found that people's attributions of consciousness to LLMs were influenced by their beliefs about the nature of consciousness and their familiarity with LLMs. Participants who were more familiar with LLMs were more likely to attribute consciousness to them, and participants who believed that consciousness is a product of complex computation were also more likely to attribute consciousness to LLMs.

Overall, the study suggests that people are generally open to the possibility that LLMs may be conscious, but they also recognize that LLMs are not as conscious as humans. These findings have implications for the development and use of LLMs, as they suggest that people may be more willing to trust and interact with LLMs that they believe are conscious.

Our evolved intuitions about privacy aren’t made for this era

Joe Green & Azim Shariff

psyche.co

Originally published September 16, 2021

Here is an excerpt:

Our concern for privacy has its evolutionary roots in the need to maintain boundaries between the self and others, for safety and security. The motivation for personal space and territoriality is a common phenomenon within the animal kingdom. Among humans, this concern about regulating physical access is complemented by one about regulating informational access. The language abilities, complex social lives and long memories of human beings made protecting our social reputations almost as important as protecting our physical bodies. Norms about sexual privacy, for instance, are common across cultures and time periods. Establishing basic seclusion for secret trysts would have allowed for all the carnal benefits without the unwelcome reputational scrutiny.

Since protection and seclusion must be balanced with interaction, our privacy concern is tuned to flexibly respond to cues in our environment, helping to determine when and what and with whom we share our physical space and personal information. We reflexively lower our voices when strange or hostile interlopers come within earshot. We experience an uneasy creepiness when someone peers over our shoulder. We viscerally feel the presence of a crowd and the public scrutiny that comes with it.

However, just as the turtles’ light-orienting reflex was confounded by the glow of urban settlements, so too have our privacy reactions been confounded by technology. Cameras and microphones – with their superhuman sensory abilities – were challenging enough. But the migration of so much of our lives online is arguably the largest environmental shift in our species’ history with regard to privacy. And our evolved privacy psychology has not caught up. Consider how most people respond to the presence of others when they are in a crowd. Humans use a host of social cues to regulate how much distance they keep between themselves and others. These include facial expression, gaze, vocal quality, posture and hand gestures. In a crowd, such cues can produce an anxiety-inducing cacophony. Moreover, our hair-trigger reputation-management system – critical to keeping us in good moral standing within our group – can drive us into a delirium of self-consciousness.

However, there is some wisdom in this anxiety. Looking into the whites of another’s eyes anchors us within the social milieu, along with all of its attendant norms and expectations. As a result, we tread carefully. Our private thoughts generally remain just that – private, conveyed only to small, trusted groups or confined to our own minds. But as ‘social networks’ suddenly switched from being small, familiar, in-person groupings to online social media platforms connecting millions of users, things changed. Untethered from recognisable social cues such as crowding and proximity, thoughts better left for a select few found their way in front of a much wider array of people, many of whom do not have our best interests at heart. Online we can feel alone and untouchable when we are neither.

Consider, too, our intuitions about what belongs to whom. Ownership can be complicated from a legal perspective but, psychologically, it is readily inferred from an early age (as anyone with young children will have realised). This is achieved through a set of heuristics that provide an intuitive ‘folk psychology’ of ownership. First possession (who first possessed an object), labour investment (who made or modified an object), and object history (information about past transfer of ownership) are all cues that people reflexively use in attributing the ownership of physical things – and consequently, the right to open, inspect or enter them.

The info is here.

Generosity pays: Selfish people have fewer children and earn less money

Eriksson, K., Vartanova, I., et al. (2020)

Journal of Personality and Social Psychology, 

118(3), 532–544. 

https://doi.org/10.1037/pspp0000213

Abstract

Does selfishness pay in the long term? Previous research has indicated that being prosocial (or otherish) rather than selfish has positive consequences for psychological well-being, physical health, and relationships. Here we instead examine the consequences for individuals’ incomes and number of children, as these are the currencies that matter most in theories that emphasize the power of self-interest, namely economics and evolutionary thinking. Drawing on both cross-sectional (Studies 1 and 2) and panel data (Studies 3 and 4), we find that prosocial individuals tend to have more children and higher income than selfish individuals. An additional survey (Study 5) of lay beliefs about how self-interest impacts income and fertility suggests one reason selfish people may persist in their behavior even though it leads to poorer outcomes: people generally expect selfish individuals to have higher incomes. Our findings have implications for lay decisions about the allocation of scarce resources, as well as for economic and evolutionary theories of human behavior.

From the General Discussion

Our findings also speak to theories of the evolutionary history of otherishness in humans. It is often assumed that evolution promotes selfishness unless group selection acts as a counter-force (Sober & Wilson, 1999), possibly combined with a punishment mechanism to offset the advantage of being selfish (Henrich & Boyd, 2001). The finding that otherishness is associated with greater fertility within populations indicates that selfishness is not necessarily advantageous in the first place. Our datasets are limited to Europe and the United States, but if the mechanisms we sketched above are correct then we should also expect a similarly positive effect of otherishness on fertility in other parts of the world.

Our results paint a more complex picture for income, compared to fertility. Whereas otherish people tended to show the largest increases in incomes over time, the majority of our studies indicated that the highest absolute levels of income were associated with moderate otherishness. There are several ways in which otherishness may influence income levels and income trajectories. As noted earlier, otherish people tend to have stronger relations and social networks, and social networks are a key source of information about job opportunities (Granovetter, 1995).

A pdf is here.

The Ordinary Concept of a Meaningful Life

Prinzing, M., De Freitas, J., & Fredrickson, B. 

(2020, May 5). 

https://doi.org/10.31234/osf.io/6sx4t

Abstract

The desire for a meaningful life is ubiquitous, yet the ordinary concept of a meaningful life is poorly understood. Across six experiments (total N = 2,539), we investigated whether third-person attributions of meaning depend on the psychological states an agent experiences (feelings of interest, engagement, and fulfillment), or on the objective conditions of their life (e.g., their effects on others). Studies 1a–b found that laypeople think subjective and objective factors contribute independently to the meaningfulness of a person’s life. Studies 2a–b found that positive mental states are thought to make a life more meaningful, even if derived from senseless activities (e.g., hand-copying the dictionary). Studies 3a–b found that agents engaged in morally bad activities are not thought to have meaningful lives, even if they feel fulfilled. In short, both an agents’ subjective mental states and objective impact on the world affect how meaningful their lives appear.

General Discussion

What, according to the ordinary concept, makes a life meaningful?  Studies1a-b found that  laypeople  think positive  mental states (interest,  engagement, fulfillment) can make an agent’s life meaningful. These studies also found that, according to lay assessments, doing something that has value for others can also make an agent’s life meaningful. These findings conflict with the predominant philosophical theories of meaning in life. These theories posit an exclusive role for either positive mental states (subjectivist theories) or objective states of an agent’s life (objectivist theories), or they require that both criteria be met (hybrid theories). In contrast, we found that laypeople think an agent’s life is meaningful when either criterion is met.This indicates that the ordinary concept of a meaningful life does not fit neatly with these three philosophical theories. Instead, they seem to be captured by what we will call the independent-additive theory: subjective factors  (positive mental states like fulfillment) and objective factors (like contribution, sensibility, and morality)each affect the meaningfulness of an agent’s life, and their effects are both independent and additive.  

We investigated the roles of sensibility and morality as plausible boundary conditions for lay attributions of meaningfulness. For sensibility, we saw somewhat mixed results. Study 2a found no evidence that a life characterized by sensible activities (wine connoisseurship) was  seen as more  meaningful than a  life characterized  by senseless  activities(rubber  band collecting). However, Study 2b, with a larger sample and wider variety of vignettes, did find such  an  effect. Nevertheless, in both  studies, fulfilling  lives were seen as  more  meaningful than  unfulfilling  ones—regardless  of  whether  that fulfillment was derived  from sensible  or senseless activities.  Hence, on the ordinary concept, sensibility contributes to meaningfulness, though  not  as  much  as  fulfillment  does. Moreover, in  alignment  with  the independent-additive theory, fulfillment maintains its additive effect, independently of sensibility.  Regarding morality, Studies 3a-b found that morally good lives were viewed as much more meaningful than morally bad ones. In fact, morally bad agents were not thought to live meaningful lives, even if those agents felt very fulfilled. In contrast, morally good agents were seen as having meaningful lives even if they didn’t feel fulfilled.Nevertheless,  though the effect of morality was larger than that of fulfillment, participants still thought that a fulfilled, immoral agent was living more meaningfully than an unfulfilled, immoral agent. Supporting the independent-additive  theory,  the additive  effect  of  fulfillment was independent  of morality.

In short, we identified four factors (fulfillment, contribution, sensibility, and morality) that seem to have independent, additive effects on third-person attributions of meaningfulness.  There  may well be more such  factors.  But  the  evidence  from  these  six experiments supports a model of third-person meaningfulness judgments that—in contrast to subjectivist,  objectivist,  and  hybrid  theories—emphasizes  independent  and  additive  factors that  contribute  to  the  meaning in a person’s life.  We  have called such a model the “independent-additive theory”.

The research is here.

Morality justifies motivated reasoning in the folk ethics of belief

Cusimano, C., & Lombrozo, T. (2020, July 20).
https://doi.org/10.31234/osf.io/7r5yb

Abstract

When faced with a dilemma between believing what is supported by an impartial assessment of the evidence (e.g., that one’s friend is guilty of a crime) and believing what would better fulfill a moral obligation (e.g., that the friend is innocent), people often believe in line with the latter. But is this how people think beliefs ought to be formed? We addressed this question across three studies and found that, across a diverse set of everyday situations, people treat moral considerations as legitimate grounds for believing propositions that are unsupported by objective, evidence-based reasoning. We further document two ways in which moral evaluations affect how people prescribe beliefs to others. First, the moral value of a belief affects the evidential threshold required to believe, such that morally good beliefs demand less evidence than morally bad beliefs. Second, people sometimes treat the moral value of a belief as an independent justification for belief, and so sometimes prescribe evidentially poor beliefs to others. Together these results show that, in the folk ethics of belief, morality can justify and demand motivated reasoning.

From the Discussion

Additionally, participants reported that moral concerns affected the standards of evidence that apply to belief, such that morally-desirable beliefs require less evidence than morally-undesirable beliefs. In Study 1, participants reported that, relative to an impartial observer with the same information, someone with a moral reason to be optimistic had a wider range of beliefs that could be considered“consistent with” and “based on” the evidence.  Critically however, the broader range of beliefs that were consistent with the same evidence were only beliefs that were more morally desirable; morally undesirable beliefs were not more consistent with the evidence. In Studies 2 and 3, participants agreed more strongly that someone who had a moral reason to adopt a desirable belief had sufficient evidence to do so compared to someone who lacked a moral reason, even though they formed the same belief on the basis of the same evidence.  Likewise, on average, participants judged that when someone adopted the morally undesirable belief, they were more often judged as having insufficient evidence for doing so relative to someone who lacked a moral reason (again, even though they formed the same belief on the basis of the same evidence).  Finally, in Study 2 (though not in Study 3), these judgments replicated using an indirect measure of evidentiary quality; namely, attributions of knowledge. In sum, these findings document that one reason people may prescribe a motivated belief to someone is because morality changes how much evidence they consider to be required to hold the belief in an evidentially-sound way.

Editor's Note: Huge implications for psychotherapy.

Naïve Normativity: The Social Foundation of Moral Cognition

Kristin Andrews
Journal of the American Philosophical Association
Volume 6, Issue 1
January 2020 , pp. 36-56

Abstract

To answer tantalizing questions such as whether animals are moral or how morality evolved, I propose starting with a somewhat less fraught question: do animals have normative cognition? Recent psychological research suggests that normative thinking, or ought-thought, begins early in human development. Recent philosophical research suggests that folk psychology is grounded in normative thought. Recent primatology research finds evidence of sophisticated cultural and social learning capacities in great apes. Drawing on these three literatures, I argue that the human variety of social cognition and moral cognition encompass the same cognitive capacities and that the nonhuman great apes may also be normative beings. To make this argument, I develop an account of animal social norms that shares key properties with Cristina Bicchieri's account of social norms but which lowers the cognitive requirements for having a social norm. I propose a set of four early developing prerequisites implicated in social cognition that make up what I call naïve normativity: the ability to identify agents, sensitivity to in-group/out-group differences, the capacity for social learning of group traditions, and responsiveness to appropriateness. I review the ape cognition literature and present preliminary empirical evidence supporting the existence of social norms and naïve normativity in great apes. While there is more empirical work to be done, I hope to have offered a framework for studying normativity in other species, and I conclude that we should be open to the possibility that normative cognition is yet another ancient cognitive endowment that is not human-unique.

The info is here.

Folk standards of sound judgment: Rationality vs. Reasonableness

Igor Grossman and others
PsyArXiv Preprints
Last edited on 10 Jan 20

Abstract

Normative theories of judgment either focus on rationality – decontextualized preference maximization, or reasonableness – the pragmatic balance of preferences and socially-conscious norms. Despite centuries of work on such concepts, a critical question appears overlooked: How do people’s intuitions and behavior align with the concepts of rationality from game theory and reasonableness from legal scholarship? We show that laypeople view rationality as abstract and preference-maximizing, simultaneously viewing reasonableness as social-context-sensitive and socially-conscious, as evidenced in spontaneous descriptions, social perceptions, and linguistic analyses of the terms in cultural products (news, soap operas, legal opinions, and Google books). Further, experiments among North Americans and Pakistani bankers, street merchants, and samples engaging in exchange (vs. market-) economy show that rationality and reasonableness lead people to different conclusions about what constitutes good judgment in Dictator Games, Commons Dilemma and Prisoner’s Dilemma: Lay rationality is reductionist and instrumental, whereas reasonableness integrates preferences with particulars and moral concerns.

The research is here.

It's not biology bro: Torture and the Misuse of Science

Shane O'Mara and John Schiemann
PsyArXiv Preprints
Last edited on December 24, 2018

Abstract

Contrary to the (in)famous line in the film Zero Dark Thirty, the CIA's torture program was not based on biology or any other science. Instead, the Bush administration and the CIA decided to use coercion immediately after the 9/11 terrorist attacks and then veneered the program's justification with a patina of pseudoscience, ignoring the actual biology of torturing human brains. We reconstruct the Bush administration’s decision-making process from released government documents, independent investigations, journalistic accounts, and memoirs to establish that the policy decision to use torture took place in the immediate aftermath of the 9/11 attacks without any investigation into its efficacy. We then present the pseudo-scientific model of torture sold to the CIA based on a loose amalgamation of methods from the old KUBARK manual, reverse-engineering of SERE training techniques, and learned helplessness theory, show why this ad hoc model amounted to pseudoscience, and then catalog what the actual science of torturing human brains – available in 2001 – reveals about the practice. We conclude with a discussion of how process of policy-making might incorporate countervailing evidence to ensure that policy problems are forestalled, via the concept of an evidence-based policy brake, which is deliberately instituted to prevent a policy going forward that is contrary to law, ethics and evidence.

The info is here.

The Essence of the Individual: The Pervasive Belief in the True Self Is an Instance of Psychological Essentialism

Andrew G. Christy, Rebecca J. Schlegel, and Andrei Cimpian
Preprint

Abstract

Eight studies (N = 2,974) were conducted to test the hypothesis that the widespread folk belief in the true self is an instance of psychological essentialism. Results supported this hypothesis. Specifically, participants’ reasoning about the true self displayed the telltale features of essentialist reasoning (immutability, discreteness, consistency, informativeness, inherence, and biological basis; Studies 1–4); participants’ endorsement of true-self beliefs correlated with individual differences in other essentialist beliefs (Study 5); and experimental manipulations of essentialist thought in domains other than the self were found to “spill over” and affect the extent to which participants endorsed true-self beliefs (Studies 6–8). These findings advance theory on the origins and functions of true-self beliefs, revealing these beliefs to be a specific instance of a broader tendency to explain phenomena in the world in terms of underlying essences.

The preprint is here.

responsibility and punishment

Katrina Sifferd interviewed by Richard Marshall
3:AM Magazine
Originally posted

Here is an excerpt:

KS: Well, for one, we won’t be able to make responsibility assessments. When you show a jury a picture of a brain lighting up in such-and-such a way it means absolutely nothing to them until somebody translates the scientific data into folk psychological terms. Expert witnesses in a trial cannot just point to a dark spot on a PET scan and sit down: the scientific data is irrelevant to the defendant’s culpability until is it translated into folk concepts that push and pull responsibility assessments in different directions. For example, an expert might note that the dark spot is a brain tumor likely to result in a severe lack of impulse control, which the jury might feel undermines attribution of the highest levels of criminal intent.

I think it is interesting that some scientific data actually seems to push responsibility assessments in both directions, or in ways unanticipated by the side offering the evidence in a criminal trial. In one high profile capital sentencing hearing, the defense offered neuroscientific evidence of psychopathy in an attempt to prove diminished capacity (and thus a mitigating factor); but instead, the jury seemed to think the data made the defendant more culpable for his actions, and sentenced him to death. Is a person whose brain shows clear signs of psychopathy less responsible because of their abnormal brain function or more responsible because their brain is abnormal (and thus they are likely to be dangerous in the future)? I think it depends on the way in which the brain is dysfunctional, and maybe the reasons why it is dysfunctional. There is a lot of important work to be done making reliable translations of neuroscientific data into folk descriptions relevant to responsibility.

(cut)

KS: Different theories of punishment seem to emphasize different aspects of our cognitive capacities as most important to culpability. Bill and I have argued that deontological accounts which postulate emotional response or empathy as crucial to moral knowledge and decision-making might be more likely to excuse all psychopaths because of their apparent lack of relevant affective data. Some deontological theorists believe that a lack of appropriate emotional response translates into a wholesale lack of legal rationality. A consequentialist theory of punishment, however, may be more likely to hold some psychopaths responsible, because it emphasizes the need for rational capacities as a means to grasp and reflect upon the consequences of action given ones goals and relevant social norms (a skill successful psychopaths may possess), and not the way one feels about these consequences.

The entire interview is here.

Experimental Approaches to Free Will: Knobe and Nahmias

Joshuan Knobe and Eddy Nahmias

Knobe and Nahmias begin with an overview of the early history and aims of experimental philosophy. Then they discuss experiments on the contrast between bypassing and throughpassing intuitions about free will (8:57); Nahmias’s “theory lite view,” according to which ordinary people have no strong views about the relation between mind and brain (17:34); whether the folk have a causal or an interventionist view of agency (24:17); the effect of descriptions of determinism on folk intuitions (32:52); and Nahmias’s work on “willusionism,” inspired by his critical view of certain popularized versions of free-will skepticism (41:47). Finally, Knobe and Nahmias consider future results that could resolve some of their disagreements (48:49) and forecast the next big steps in experimental philosophy of free will (57:00).

Person as Scientist, Person as Moralist

By Joshua Knobe
Upcoming in Brain and Brain Sciences

Abstract

It has often been suggested that people’s ordinary capacities for folk psychology and causal cognition make use of much the same methods one might find in a formal scientific investigation. A series of recent experimental results offer a challenge to this widely-held view, suggesting that people’s moral judgments can influence the intuitions they hold both in folk psychology and in moral cognition. The present target article argues that these effects are best explained on a model according to which moral considerations actually figure in the fundamental competencies people use to make sense of the world.

-----------

Consider the way research is conducted in a typical modern university. There are departments for theology, drama, philosophy… and then there are departments specifically devoted to the practice of science. Faculty members in these science departments generally have quite specific responsibilities. They are not supposed to make use of all the various methods and approaches one finds in other parts of the university.

Now consider the way the human mind ordinarily makes sense of the world. One plausible view would be that the human mind works something like a modern university.   There are psychological processes devoted to religion (the mind’s theology department), to aesthetics (the mind’s art department), to morality (the mind’s philosophy department) … and then there are processes specifically devoted to questions that have a roughly ‘scientific’ character. These processes work quite differently from the ones we use in thinking about, say, moral or aesthetic questions. They proceed using more or less the
same sorts of methods we find in university science departments.

This metaphor is a powerful one, and it has shaped research programs in many different areas of cognitive science. Take the study of folk psychology. Ordinary people have a capacity to ascribe mental states (beliefs, desires, etc.), and researchers have sometimes suggested that people acquire this capacity in much the same way that scientists develop theoretical frameworks (e.g., Gopnik & Wellman 1992). Or take causal cognition. Ordinary people have an ability to determine whether one event caused another, and it has been suggested that they do so by looking at the same sorts of statistical information scientists normally consult (e.g., Kelley 1967). Numerous other fields have taken a similar path. In each case, the basic strategy is to look at the methods used by professional research scientists and then to hypothesize that people actually use similar methods in their ordinary understanding. This strategy has clearly led to many important advances.

The entire article is here.