Theory Is All You Need: AI, Human Cognition, and Causal Reasoning

Felin, T., & Holweg, M. (2024).

SSRN Electronic Journal.

https://doi.org/10.2139/ssrn.4737265

Abstract

Scholars argue that artificial intelligence (AI) can generate genuine novelty and new knowledge and, in turn, that AI and computational models of cognition will replace human decision making under uncertainty. We disagree. We argue that AI’s data-based prediction is different from human theory-based causal logic and reasoning. We highlight problems with the decades-old analogy between computers and minds as input–output devices, using large language models as an example. Human cognition is better conceptualized as a form of theory-based causal reasoning rather than AI’s emphasis on information processing and data-based prediction. AI uses a probability-based approach to knowledge and is largely backward looking and imitative, whereas human cognition is forward-looking and capable of generating genuine novelty. We introduce the idea of data–belief asymmetries to highlight the difference between AI and human cognition, using the example of heavier-than-air flight to illustrate our arguments. Theory-based causal reasoning provides a cognitive mechanism for humans to intervene in the world and to engage in directed experimentation to generate new data. Throughout the article, we discuss the implications of our argument for understanding the origins of novelty, new knowledge, and decision making under uncertainty.

Here are some thoughts:

This paper challenges the dominant view that artificial intelligence (AI), particularly large language models (LLMs), mirrors or will soon surpass human cognition. The authors argue against the widespread computational metaphor of the mind, which treats human thinking as data-driven, predictive information processing akin to AI. Instead, they emphasize that human cognition is fundamentally theory-driven and rooted in causal reasoning, experimentation, and the generation of novel, heterogenous beliefs—often in defiance of existing data or consensus. Drawing on historical examples like the Wright brothers, who succeeded despite prevailing scientific skepticism, the paper illustrates how human progress often stems from delusional-seeming ideas that later prove correct. Unlike AI systems that rely on statistical pattern recognition and next-word prediction from vast datasets, humans engage in counterfactual thinking, intentional intervention, and theory-building, enabling true innovation and scientific discovery. The authors caution against over-reliance on prediction-based AI in decision-making, especially under uncertainty, and advocate for a "theory-based view" of cognition that prioritizes causal understanding over mere correlation. In essence, they contend that while AI excels at extrapolating from the past, only human theory-making can generate genuinely new knowledge.

Latent motives guide structure learning during adaptive social choice

van Baar, J.M., Nassar, M.R., Deng, W. et al.

Nat Hum Behav (2021). 

https://doi.org/10.1038/s41562-021-01207-4

Abstract

Predicting the behaviour of others is an essential part of social cognition. Despite its ubiquity, social prediction poses a poorly understood generalization problem: we cannot assume that others will repeat past behaviour in new settings or that their future actions are entirely unrelated to the past. We demonstrate that humans solve this challenge using a structure learning mechanism that uncovers other people’s latent, unobservable motives, such as greed and risk aversion. In four studies, participants (N = 501) predicted other players’ decisions across four economic games, each with different social tensions (for example, Prisoner’s Dilemma and Stag Hunt). Participants achieved accurate social prediction by learning the stable motivational structure underlying a player’s changing actions across games. This motive-based abstraction enabled participants to attend to information diagnostic of the player’s next move and disregard irrelevant contextual cues. Participants who successfully learned another’s motives were more strategic in a subsequent competitive interaction with that player in entirely new contexts, reflecting that social structure learning supports adaptive social behaviour.

Significance statement

A hallmark of human cognition is being able to predict the behavior of others. How do we achieve social prediction given that we routinely encounter others in a dizzying array of social situations? We find people achieve accurate social prediction by inferring another’s hidden motives—motives that do not necessarily have a one-to-one correspondence with observable behaviors. Participants were able to infer another’s motives using a structure learning mechanism that enabled generalization.  Individuals used what they learned about others in one setting to predict their actions in an entirely new setting. This cognitive process can explain a wealth of social behaviors, ranging from strategic economic decisions to stereotyping and racial bias.

From the Discussion

How do people construct and apply abstracted mental models of others’ motives? Our data suggest that attention plays a key role in guiding this process. Attention is a fundamental cognitive mechanism as it affords optimal access to behaviorally relevant information with limited processing capacity. Our findings show how attention supports social prediction. In the Social Prediction Game, as in everyday social interactions, there were multiple cues that could be predictive of another’s behavior, from the player payoffs S and T to the order of the games or even the initials of the player. Structure learning allowed participants to disregard superficial cues and attend to information relevant to the players’ latent motives. Although this process facilitated accurate social prediction with limited effort if the inferred motives were correct, incorrect structure learning caused counterproductive attention on irrelevant information. For example, participants who did not consider risk aversion failed to shift their attention to the sucker’s payoff (S) during the Pessimist block and instead kept looking at the temptation to defect (T), thereby missing out on information predictive of the player’s choices. This suggests that what we can learn about other people is limited by our expectations.

A pdf of the research is here.

Neanderthals And Humans Were at War For Over 100,000 Years, Evidence Shows

Nicholas Longrich

The Conversation

Originally posted 3 Nov 20

Here is an excerpt:

Why else would we take so long to leave Africa? Not because the environment was hostile but because Neanderthals were already thriving in Europe and Asia.

It's exceedingly unlikely that modern humans met the Neanderthals and decided to just live and let live. If nothing else, population growth inevitably forces humans to acquire more land, to ensure sufficient territory to hunt and forage food for their children.

But an aggressive military strategy is also good evolutionary strategy.

Instead, for thousands of years, we must have tested their fighters, and for thousands of years, we kept losing. In weapons, tactics, strategy, we were fairly evenly matched.

Neanderthals probably had tactical and strategic advantages. They'd occupied the Middle East for millennia, doubtless gaining intimate knowledge of the terrain, the seasons, how to live off the native plants and animals.

In battle, their massive, muscular builds must have made them devastating fighters in close-quarters combat. Their huge eyes likely gave Neanderthals superior low-light vision, letting them manoeuvre in the dark for ambushes and dawn raids.

Sapiens victorious

Finally, the stalemate broke, and the tide shifted. We don't know why. It's possible the invention of superior ranged weapons – bows, spear-throwers, throwing clubs – let lightly-built Homo sapiens harass the stocky Neanderthals from a distance using hit-and-run tactics.

Or perhaps better hunting and gathering techniques let sapiens feed bigger tribes, creating numerical superiority in battle.

Even after primitive Homo sapiens broke out of Africa 200,000 years ago, it took over 150,000 years to conquer Neanderthal lands. In Israel and Greece, archaic Homo sapiens took ground only to fall back against Neanderthal counteroffensives, before a final offensive by modern Homo sapiens, starting 125,000 years ago, eliminated them.

The info is here.

4 Ways Lying Becomes the Norm at a Company

Ron Carucci
Harvard Business Review
Originally posted February 15, 2019

Here is an excerpt:

Unjust accountability systems. When an organization’s processes for measuring employee contributions is perceived as unfair or unjust, we found it is 3.77 times more likely to have people withhold or distort information. We intentionally excluded compensation in our research, because incentive structures can sometimes play disproportionate roles in influencing behavior, and simply looked at how contribution was measured and evaluated through performance management systems, routine feedback processes, and cultural recognition. One interviewee captured a pervasive sentiment about how destructive these systems can be: “I don’t know why I work so hard. My boss doesn’t have a clue what I do. I fill out the appraisal forms at the end of the year, he signs them and sends them to HR. We pretend to have a discussion, and then we start over. It’s a rigged system.” Our study showed that when accountability processes are seen as unfair, people feel forced to embellish their accomplishments and hide, or make excuses for their shortfalls. That sets the stage for dishonest behavior. Research on organizational injustice shows a direct correlation between an employee’s sense of fairness and a conscious choice to sabotage the organization. And more recent research confirms that unfair comparison among employees leads directly to unethical behavior.

Fortunately, our statistical models show that even a 20% improvement in performance management consistency, as evidenced by employees belief that their contributions have been fairly assessed against known standards, can improve truth telling behavior by 12%.

The info is here.