Human-like intuitive behavior and reasoning biases emerged in large language models but disappeared in ChatGPT.

We design a battery of semantic illusions and cognitive reflection tests, aimed to elicit intuitive yet erroneous responses. We administer these tasks, traditionally used to study reasoning and decision-making in humans, to OpenAI's generative pre-trained transformer model family. The results show that as the models expand in size and linguistic proficiency they increasingly display human-like intuitive system 1 thinking and associated cognitive errors. This pattern shifts notably with the introduction of ChatGPT models, which tend to respond correctly, avoiding the traps embedded in the tasks. Both ChatGPT-3.5 and 4 utilize the input-output context window to engage in chain-of-thought reasoning, reminiscent of how people use notepads to support their system 2 thinking. Yet, they remain accurate even when prevented from engaging in chain-of-thought reasoning, indicating that their system-1-like next-word generation processes are more accurate than those of older models. Our findings highlight the value of applying psychological methodologies to study large language models, as this can uncover previously undetected emergent characteristics.

Empathic concern and personal distress depend on situational but not dispositional factors.

Empathic concern and personal distress are empathic responses that may result when observing someone in discomfort. Even though these empathic responses have received much attention in past research, it is still unclear which conditions contribute to their respective experience. Hence, the main goal of this study was to examine if dispositional empathic traits or rather situational variables are more likely to evoke empathic concern and personal distress and how the two empathic responses influence motor responses. We presented pictures of persons in psychological, physical, or no pain with matched descriptions of situations that promoted an other-focused state. Approach-avoidance movements were demanded by a subsequently presented tone. While psychological pain led to more empathic concern, physical pain led to higher ratings of personal distress. Linear mixed-effects modelling analysis further revealed that situational factors, such as the type of pain but also the affect experienced by the participants before the experiment predicted the two empathic responses, whereas dispositional empathic traits had no significant influence. In addition, the more intensely the empathic responses were experienced, the faster were movements initiated, presumably reflecting an effect of arousal. Overall, the present study advances our understanding of empathic responses to people in need and provides novel methodological tools to effectively manipulate and analyze empathic concern and personal distress in future research.

Racial bias in empathy: Do we process dark- and fair-colored hands in pain differently? An EEG study.

The aim of this study was to identify racial bias influences on empathic processing from early stimulus encoding, over categorization until late motor processing stages by comparing brain responses (electroencephalogram) to pictures of fair- and dark-colored hands in painful or neutral daily-life situations. Participants performed a pain judgment task and a skin color judgment task. Event-related brain potentials (ERPs) substantiated former findings of automatic empathic influences on stimulus encoding, reflected by the early posterior negativity (EPN), and late controlled influences on the stimulus categorization, as reflected by the late posterior positivity (P3b). Concerning the racial bias in empathy (RBE) effect, more positive amplitudes in the 280-340 ms time window over frontocentral electrodes in the painful than the neutral condition for fair- but not dark-colored hands speak for an early influence of racial bias. This was further supported by correlations with empathic concern scores for fair- but not dark-colored stimuli. Additionally, P3b amplitude differences between fair- and dark-colored hands to painful stimuli increased with the implicit racial attitude of participants, suggesting that the categorization stage is not completely immune to racial bias. Regarding the motor processing stages, power change values in the upper beta-band (19-30 Hz) revealed for painful compared to neutral stimuli larger facilitation of sensorimotor activity before the response and larger inhibition after the response, independent of skin color. In conclusion, present findings speak for an influence of the RBE on early perceptual encoding but also on the late categorization stage that depends on the participant's implicit attitude towards racial outgroups.

Empathy for pain influences perceptual and motor processing: Evidence from response force, ERPs, and EEG oscillations.

In the present study we investigated the nature and chronometry of empathy for pain influences on perceptual and motor processes. Thus, event-related brain potentials (ERPs), response force (RF) and oscillatory electroencephalography (EEG) activity were measured while participants were presented with pictures of body parts in painful or neutral situations. Their task consisted in either judging the painfulness of the stimuli or counting the body parts displayed. ERP results supported the assumption of an early automatic component of empathy for pain, as reflected by the early posterior negativity (EPN), and of a late controlled component, as reflected by the late posterior positivity (P3). RF indicated that empathy-evoking stimuli facilitate motor responses if attention is directed toward the pain dimension, whereas EEG oscillations in the mu-and beta-band revealed, independent of the task, an enhanced activation of the sensorimotor cortex after the response to painful compared to neutral stimuli. In conclusion, present findings indicate that empathy-evoking stimuli produce automatic and controlled effects on both perceptual and motor processing.