Predation problem

From Wikipedia, the free encyclopedia

https://en.wikipedia.org/wiki/Predation_problem 

A snowy owl carries a killed American black duck

The predation problem or predation argument refers to the consideration of the harms experienced by animals due to predation as a moral problem, that humans may or may not have an obligation to work towards preventing. Discourse on this topic has, by and large, been held within the disciplines of animal and environmental ethics. The issue has particularly been discussed in relation to animal rights and wild animal suffering. Some critics have considered an obligation to prevent predation as untenable or absurd and have used the position as a reductio ad absurdum to reject the concept of animal rights altogether. Others have criticized any obligation implied by the animal rights position as environmentally harmful.

Responses from animal ethicists and rights advocates have been varied. Some have rejected the claim that animal rights as a position implies that we are obligated to prevent predation, while others have argued that the animal rights position does imply that predation is something that we should try to avert. Others have asserted that it is not something that we should do anything about now due to the risk that we could inadvertently cause significant harm, but that it is something that we may be able to effectively take action on in the future with improved knowledge and technologies.

Historical views

Problem of evil

See also: Problem of evil § Problem of evil and animal suffering, and Evolutionary theodicy

Predation has historically been viewed as a natural evil within the context of the problem of evil and has been considered a moral concern for Christians who have engaged with theodicy. Natural evils have been sometimes thought of as something that humans should work towards alleviating, or as part of a greater good which justifies the existence of this type of evil. Thomas Aquinas advocated the latter view, arguing that "defects" in nature such as predation led to the "good of another, or even to the universal good" and that if "all evil were prevented, much good would be absent from the universe". Within Christian and Hebrew Scripture, there are several prophecies which describe a future Heaven or Earth where predation is no longer a feature of nature, including Isaiah's prophecy that "[t]he wolf shall live with the lamb, the leopard shall lie down with the kid, the calf and the lion and the fatling together, and a little child shall lead them."

In his notebooks (written between 1487 and 1505), Leonardo da Vinci suggested that natural suffering and death, including plagues and predation, are necessary for maintaining balance and renewal in the world, even if they seem unjust or cruel. David Hume made several observations about predation and suffering experienced by wild animals in Dialogues Concerning Natural Religion (1779), stating that the "stronger prey upon the weaker, and keep them in perpetual terror and anxiety".

William Paley, in Natural Theology, described predation as being the most challenging of God's work to establish the utility of, nevertheless, he defended predation as the means to deal with the potentially catastrophic effects of animals producing more offspring than can possibly survive.

The debate around predation and the problem of evil was significantly increased by the popularization of Charles Darwin's theory of natural selection. Some earlier Christians argued that violence in nature was a result of the fall of man, but evidence that predation has existed for millions of years before the evolution of humans and the concept of sin, indicates that while life has existed, there has never been a time when nature has been free from violence. Darwin himself questioned how the fact that the Ichneumonidae prey on the bodies of living caterpillars could be reconciled with the idea of an omnibenevolent God.

Criticism of moral judgements towards predatory animals

Plutarch criticised the labelling of carnivorous animals such as lions, tigers and snakes as barbarous because for them killing is a necessity while for humans who can live off of "nature's beneficent fruits" killing is a "luxury and crime".

The writer Edward Augustus Kendall discussed predation in his book of moral fables The Canary Bird (1799), in which he argued that predatory behavior by animals should not be judged by human moral standards and that "a prejudice against particular creatures, for fancied acts of cruelty is absurd".

Philosophical pessimism

Giacomo Leopardi, the Italian poet and philosopher, in Operette morali (1827) engaged in a dialogue with Nature in "Dialogue between Nature and an Icelander", which uses the inevitability of predation—such as a squirrel fleeing from a rattlesnake, only to run into the snake's open mouth—as a moral indictment on nature's cannibalism of its own offspring. The inevitability of such cycles of destruction and creation was a cause for Leopardi's philosophical pessimism. In Zibaldone, published posthumously in 1898, Leopardi argued that predation is the ultimate indication of the evil design of nature.

Similar to Leopardi, the German philosopher Arthur Schopenhauer, in 1851, used the pain experienced by an animal being devoured by another as a refutation against the idea that the "pleasure in the world outweighs the pain".

Animal rights

Lewis Gompertz, an early animal rights advocate, and one of the first contemporary authors to address the problem of wild animal suffering, in the fifth chapter of his 1824 book Moral Inquiries on the Situation of Man and of Brutes, engaged in a dialogue, in which he asserted that animals devouring each other can be judged as wrong by the rules that we use to govern human lives and stated that "should I witness the attempt in any animal of destroying another, I would endeavour to frustrate it; though this might probably be wrong." He went on to argue that the extinction of carnivorous species would not be bad, claiming that the species of one animal is not more important than an equal number of another and that it would be possible for some carnivorous animals, like wolves, to instead sustain themselves on vegetables.

The American zoologist and animal rights philosopher J. Howard Moore in the pamphlet Why I Am a Vegetarian, published in 1895, described the carnivora as "relentless brutes", whose existence is a travesty for ethics, justice and mercy. In Better-World Philosophy (1899), Moore argued that carnivorousness was the result of excessive egoism, a product of natural selection, stating "Life riots on life—tooth and talon, beak and paw". He went on to claim that the irredeemable nature of carnivorous species meant that they could not be reconciled with each other in his ideal arrangement of the universe, which he called a "Confederation of the Consciousnesses". In The New Ethics (1907), Moore labelled carnivorous species as "criminal" races whose "existence is a continual menace to the peace and well-being of the world" because the "fullness of their lives is dependent upon the emptiness and destruction of others".

In 1903, the Scottish philosopher David G. Ritchie in response to Henry S. Salt's 1892 book Animals' Rights, claimed that giving animals rights would imply that we must "protect the weak among them against the strong" and to achieve this, carnivorous animals should be put to death or slowly starved by "permanent captivity and vegetarian diet". He considered this proposal absurd, stating that the "declaration of the rights of every creeping thing [is] to remain a mere hypocritical formula to gratify pug-loving sentimentalists".

Contemporary views

Animal ethics

In 1973, Australian philosopher Peter Singer argued that if humans were to try to prevent predation, such as from stopping lions killing gazelles, that it would likely increase the "net amount of animal suffering", but asserted that if hypothetically we could reduce suffering in the long-term, then it would be right to intervene.

The English philosopher Stephen R. L. Clark's "The Rights of Wild Things" (1979) is considered to be one of the first ethics papers to explicitly engage with predation as a problem. In the paper, Clark argues that the concept that humans are obligated to aid animals against predators is not absurd, but that it follows only in the abstract, not in practice.

Animal rights philosopher, Tom Regan in his 1983 book, The Case for Animal Rights, argued that humans have no obligation to prevent predation because carnivorous animals are not moral agents and as a result cannot violate the rights of the animals that they predate. Along these lines, Julius Kapembwa argues that "intervention in predation is neither required nor permitted by animal rights theory".

Steve Sapontzis, in his 1984 paper "Predation" argues against the idea that the problem of predation is a reductio ad absurdum for animal rights, instead, he claims that if we accept the view that we have an obligation to reduce avoidable animal suffering, then predation is something that we should work towards preventing if we can do so without inflicting greater suffering. Sapontzis concludes that whether humans choose to fulfil this particular obligation, or attempt to reduce other forms of avoidable suffering, is a question of where humans can do the most good.

In a 2003 paper, the economist Tyler Cowen advocates, from a utility, rights and holistic perspective, for the policing of nature to reduce the predatory activity of certain animals to help their victims.

The transhumanist philosopher David Pearce, in his 2009 essay, "Reprogramming Predators", claims that predation is an immense source of suffering in the world and that a "biosphere without suffering is technically feasible". He argues for the phased extinction of carnivorous species using immunocontraceptives or "reprogramming" them using gene editing so that their descendants become herbivores. Pearce lists and argues against a number of justifications used by people who think that suffering caused by predation does not matter and that it should be conserved in its current state, including a "television-based conception of the living world", "[s]elective realism" and "[a]daptive empathy deficits".

In 2010, Jeff McMahan published "The Meat Eaters", an op-ed for the New York Times on predation as a moral issue, in which he argued that preventing the massive amounts of suffering and death caused by predation would be a good thing and that the extinction of carnivorous species could be instrumentally good if this could be achieved without inflicting "ecological upheaval involving more harm than would be prevented by the end of predation". McMahan received a number of objections to his arguments and responded to these in another op-ed published in the same year, "Predators: A Response". He later published his arguments as a chapter titled "The Moral Problem of Predation", in the 2015 book Philosophy Comes to Dinner.

Peter Vallentyne argues that it is permissible for humans to intervene to help prey in limited ways, if the cost to humans is minimal, but that we should not eliminate predators. In the same way that we aid humans in need, when the cost to humans is minimal, humans might help wild animals in limited circumstances.

Martha Nussbaum asserts that the predation problem and what should be done to solve it should be the subject of serious discussion, also arguing that there should be research into future solutions. Nussbaum draws attention to a need to convince people that predation is a problem and to challenge the common conception of predation as exciting and enthralling, which she believes has a negative impact on human culture. She goes on to challenge the idea of animals, who are preyed upon, as existing to be food for other animals, rather than being made to live for their own lives. Nussbaum concludes that humans, who have extensive control over animal lives and habitats, need to face up to their responsibilities towards wild animals and work towards their flourishing, rather than harming them.

Some ethicists have made concrete proposals for reducing or preventing predation, including stopping the reintroduction of predators in locations where they have previously gone extinct, and removing predators from wild areas.

Environmental ethics

In 1984, the British ecologist Felicity A. Huntingford published "Some ethical issues raised by studies of predation and aggression", in which she discusses ethical issues and implications regarding the staging of artificial encounters for studies of predator-prey interactions.

In the context of ecology, predation is widely regarded as playing an important and necessary role in ecosystems. This has led some writers, such as Michael Pollan, to reject predation as being a moral problem at all, stating "predation is not a matter of morality or politics; it, also, is a matter of symbiosis". Under Aldo Leopold's land ethic, native predators, as components of biotic communities, are considered important to conserve.

The environmental philosopher J. Baird Callicott asserts that the implication of animal rights theory, namely that we should protect animals from predators, would "[n]ot only [result in] the (humane) eradication of predators destroy the community, it would destroy the species which are the intended beneficiaries of this misplaced morality. Many prey species depend upon predators to optimize their populations." Holmes Rolston III views predation as an essential natural process and driver of evolution, that is a "sad good" to be respected and valued. Ty Raterman, an environmentalist, has argued that predation is something that can be lamented without implying that we have an obligation to prevent it.

The environmental ethicist William Lynn has argued that from a welfare perspective predation "is necessary for the well-being of predators and prey" and essential for the maintenance of the integrity of the ecological communities. Larry Rasmussen, a Christian environmental ethicist, has argued that predation is "not a pattern of morality we praise and advocate".

Other uses of the term

"Predation problem" can also refer to the predation of animals who belong to species considered valuable to humans for economic reasons or conservation, such as domestic sheep predation by coyotes, farmed salmon predation by seals, the predation of animals who are hunted for sport or food and cat predation of wild animals; culling or removal of predatory animals may be carried out to reduce such incidents.

Artificial intelligence in mental health

From Wikipedia, the free encyclopedia

https://en.wikipedia.org/wiki/Artificial_intelligence_in_mental_health 

Artificial intelligence in mental health refers to the application of artificial intelligence (AI), computational technologies and algorithms to support the understanding, diagnosis, and treatment of mental health disorders. In the context of mental health, AI is considered a component of digital healthcare, with the objective of improving accessibility and accuracy and addressing the growing prevalence of mental health concerns. Applications of AI in this field include the identification and diagnosis of mental disorders, analysis of electronic health records, development of personalized treatment plans, and analytics for suicide prevention. There is also research into, and private companies offering, AI therapists that provide talk therapies such as cognitive behavioral therapy. Despite its many potential benefits, the implementation of AI in mental healthcare presents significant challenges and ethical considerations, and its adoption remains limited as researchers and practitioners work to address existing barriers. There are concerns over data privacy and training data diversity.

Background

In 2019, 1 in every 8 people, or 970 million people around the world were living with a mental disorder, with anxiety and depressive disorders being the most common. In 2020, the number of people living with anxiety and depressive disorders rose significantly because of the COVID-19 pandemic. Additionally, the prevalence of mental health and addiction disorders exhibits a nearly equal distribution across genders, emphasizing the widespread nature of the issue.

The use of AI in mental health aims to support responsive and sustainable interventions against the global challenge posed by mental health disorders. Some issues common to the mental health industry are provider shortages, inefficient diagnoses, and ineffective treatments. The global market for AI-driven mental health applications is projected to grow significantly, with estimates suggesting an increase from US$0.92 billion in 2023 to US$14.89 billion by 2033. This growth indicates a growing interest in AI's ability to address critical challenges in mental healthcare provision through the development and implementation of innovative solutions.

AI-driven approaches

Several AI technologies, including machine learning (ML), natural language processing (NLP), deep learning (DL), computer vision (CV) and LLMs and generative AI are currently applied in various mental health contexts. These technologies enable early detection of mental health conditions, personalized treatment recommendations, and real-time monitoring of patient well-being.

Machine learning

Machine learning is an AI technique that enables computers to identify patterns in large datasets and make predictions based on those patterns. Unlike traditional medical research, which begins with a hypothesis, ML models analyze existing data to uncover correlations and develop predictive algorithms. ML in psychiatry is limited by data availability and quality. Many psychiatric diagnoses rely on subjective assessments, interviews, and behavioral observations, making structured data collection difficult. Some researchers have applied transfer learning, a technique that adapts ML models trained in other fields, to overcome these challenges in mental health applications.

Deep learning

Deep learning, a subset of ML, involves neural networks with many layers of neurons, that can grasp complex patterns, similarly to human brains. It is particularly useful for identifying subtle patterns in speech, imaging, and physiological data. Deep learning techniques have been applied in neuroimaging research to identify abnormalities in brain scans associated with conditions such as schizophrenia, depression, and PTSD. However, deep learning models require extensive, high-quality datasets to function effectively. The limited availability of large, diverse mental health datasets poses a challenge, as patient privacy regulations restrict access to medical records. Additionally, deep learning models often operate as "black boxes", meaning their decision-making processes are not easily interpretable by clinicians, raising concerns about transparency and clinical trust.

Natural language processing

Natural language processing allows AI systems to analyze and interpret human language, including speech, text, and tone of voice. In mental health, NLP is used to extract meaningful insights from conversations, clinical notes, and patient-reported symptoms. NLP can assess sentiment, speech patterns, and linguistic cues to detect signs of mental distress. This is crucial because many of the diagnoses and DSM-5 mental health disorders are diagnosed via speech in doctor-patient interviews, utilizing the clinician's skill for behavioral pattern recognition and translating it into medically relevant information to be documented and used for diagnoses. As research continues, NLP models must address ethical concerns related to patient privacy, consent, and potential biases in language interpretation.

Advancements in NLP such as sentiment analysis identifies distinctions in tone and speech to detect anxiety and depression. "Woebot", uses sentiment analysis to scrutinize and detect patterns for depression or despair and suggests professional help to patients. Similarly, "Cogito", an AI platform uses voice analysis to find changes in pitch and loudness to identify symptoms of depression or anxiety. The application of NLP can contribute to early diagnosis and improved treatment strategies.

Computer vision

Computer vision enables AI to analyze visual data, such as facial expressions, body language, and micro expressions, to assess emotional and psychological states. This technology is increasingly used in mental health research to detect signs of depression, anxiety, and PTSD through facial analysis. Computer vision tools have been explored for their ability to detect nonverbal cues, such as hesitation or changes in eye contact, which may correlate with emotional distress. Despite its potential, computer vision in mental health raises ethical and accuracy concerns. Facial recognition algorithms can be influenced by cultural and racial biases, leading to potential misinterpretations of emotional expressions. Additionally, concerns about informed consent and data privacy must be addressed before widespread clinical adoption.

LLMs and generative AI

Research studies and social media posts indicate that some individuals seek therapeutic or emotional support from LLMs. A survey in early 2025 by Sentio University found that 48.7 percent of 499 U.S. adults with self-reported mental health conditions who used LLMs had turned to them for support with anxiety, depression, loneliness, or related issues. LLMs can offer lower-cost and increased accessibility compared to traditional mental health services. LLMs are known to generate hallucinations, which are plausible but inaccurate statements that may mislead users in sensitive contexts. Additional research has found that LLMs can display stigmatizing responses or inappropriately validate maladaptive thoughts, underscoring limits in replicating the judgment and relational capacities of trained clinicians. Crisis evaluations suggest that some systems do not consistently perform essential safety tasks, including suicide risk assessment or referral to appropriate services. Research on empathy expressed by LLMs is mixed, with a systematic review reporting that in some studies their responses are rated as more empathic than those of clinicians, and other work in medical ethics warning that such systems lack genuine emotional intelligence and can reproduce inequities in health care.

Applications

Diagnosis

AI with the use of NLP and ML can be used to help diagnose individuals with mental health disorders. It can be used to differentiate closely similar disorders based on their initial presentation to inform timely treatment before disease progression. For example, it may be able to differentiate unipolar from bipolar depression by analyzing imaging and medical scans. AI can examine different biomarkers to help determine not only the disorder a patient may have, but the type and level of care needed as well. AI also has the potential to identify novel diseases that were overlooked due to the heterogeneity of presentation of a single disorder. Doctors may overlook the presentation of a disorder because while many people get diagnosed with depression, that depression may take on different forms and be enacted in different behaviors. AI can parse through the variability found in human expression data and potentially identify different types of depression.

Prognosis

AI can be used to create accurate predictions for disease progression once diagnosed. AI algorithms can also use data-driven approaches to build new clinical risk prediction models without relying primarily on current theories of psychopathology. However, internal and external validation of an AI algorithm is essential for its clinical utility. In fact, some studies have used neuroimaging, electronic health records, genetic data, and speech data to predict how depression would present in patients, their risk for suicidality or substance abuse, or functional outcomes. The prognosis seems to be highly promising, though it comes with important challenges and ethical considerations such as:

Early detention AI can analyze patterns in speech, writing, facial expressions, and social media behavior to detect early signs of depression, anxiety, PTSD, and even schizophrenia.

Treatment

In psychiatry, in many cases multiple drugs are trialed with the patients until the correct combination or regimen is reached to effectively treat their ailment—AI systems have been investigated for their potential to predict treatment response based on observed data collected from various sources. This application of AI has the potential to reduce the time, effort, and resources required while alleviating the burden on both patients and clinicians.

Benefits

Artificial intelligence offers several potential advantages in the field of mental health care:

• Enhanced diagnostic accuracy: AI systems are capable of analyzing large datasets including brain imaging, genetic testing, and behavioral data to detect biomarkers associated with mental health conditions. This may contribute to more accurate and timely diagnoses.
• Personalized treatment planning: AI algorithms can process information from electronic health records (EHRs), neuroimaging, and genomic data to identify the most effective treatment strategies tailored to individual patients.
• Improved access to care: AI technologies can facilitate the delivery of mental health services such as cognitive behavioral therapy (CBT) through virtual platforms. This may increase access to care, particularly in underserved or remote areas.
• Early detection and monitoring: AI tools can assist clinicians in recognizing early warning signs of mental health disorders, enabling proactive interventions and potentially reducing the risk of acute episodes or hospitalizations.
• Use of chatbots and virtual assistants: AI-powered systems can support administrative functions, including appointment scheduling, patient triage, and organizing medical history. This may improve operational efficiency and enhance patient engagement.
• Predictive analytics for suicide prevention: AI models can analyze behavioral, clinical, and social data to identify individuals at elevated risk of suicide, enabling targeted prevention strategies and informing public health policies.

Challenges

Despite its potential, the application of AI in mental health presents a number of ethical, practical, and technical challenges:

• Informed consent and transparency: The complexity and opacity of AI systems particularly in how they process data and generate outputs require clinicians to clearly communicate potential limitations, biases, and uncertainties to patients as part of the informed consent process.
• Right to explanation: Patients may request explanations regarding AI-generated diagnoses or treatment recommendations. Healthcare providers have a responsibility to ensure that these explanations are available and comprehensible.
• Privacy and data protection: The use of AI in mental health care must balance data utility with the protection of sensitive personal information. Ensuring robust privacy safeguards is essential to building trust among users.
• Lack of diversity in training data: AI models often rely on datasets that may not be representative of diverse populations. This can lead to biased outcomes and reduced effectiveness in diagnosing or treating individuals from underrepresented groups.
• Provider skepticism and implementation barriers: Clinicians and health care organizations may be hesitant to adopt AI tools due to a lack of familiarity, concerns about reliability, or uncertainty about integration into existing care workflows.
• Responsibility and the "Tarasoff duty": In cases where AI identifies a patient as a potential risk to themselves or others, it remains unclear who holds the legal and ethical responsibility to act particularly in jurisdictions with mandatory duty-to-warn obligations.
• Data quality and accessibility: High-quality mental health data is often difficult to obtain due to ethical constraints and privacy concerns. Limited access to diverse and comprehensive datasets may hinder the accuracy and real-world applicability of AI systems.
• Bias in data: Bias in data algorithms means placing preferences of certain groups of people over others which is unfair. AI models are constructed with such biases leading to wrong treatment, incorrect diagnoses and harmful medical outcomes. Because of such bias, groups from diverse backgrounds could be at risk of being underrepresented. Most AI systems are trained on western populations data that can also be a cause of algorithmic bias. If AI systems cannot be trained on inclusive data, it risks increasing racial disparities and mental health issues.

Current AI trends in mental health

As of 2020, the Food and Drug Administration (FDA) had not yet approved any artificial intelligence-based tools for use in psychiatry. However, in 2022, the FDA granted authorization for the initial testing of an AI-driven mental health assessment tool known as the AI-Generated Clinical Outcome Assessment (AI-COA). This system employs multimodal behavioral signal processing and machine learning to track mental health symptoms and assess the severity of anxiety and depression. AI-COA was incorporated into a pilot program to evaluate its clinical effectiveness. As of 2025, it has not received full regulatory approval.

Mental health tech startups

Mental health tech startups continue to lead investment activity in digital health despite the ongoing impacts of macroeconomic factors like inflation, supply chain disruptions, and interest rates.

According to CB Insights, State of Mental Health Tech 2021 Report, mental health tech companies raised $5.5 billion worldwide (324 deals), a 139% increase from the previous year that recorded 258 deals.

A number of startups that are using AI in mental healthcare have closed notable deals in 2022 as well. Among them is the AI chatbot Wysa ($20 million in funding), BlueSkeye that is working on improving early diagnosis (£3.4 million), the Upheal smart notebook for mental health professionals ($10 million in funding), and the AI-based mental health companion clare&me (€1 million). Founded in 2021, Earkick serves as an 'AI therapist' for mental health support.

Alongside patient-facing applications, clinician-facing AI platforms have also emerged to support mental healthcare delivery. These tools are designed to assist practitioners with tasks such as documentation and workflow management rather than providing direct therapy. One example is Heidi Health, an AI-assisted clinical documentation system used by mental health practitioners to support the creation of structured clinical notes.

Emotional AI and predictive detection

An analysis of the investment landscape and ongoing research suggests that we are likely to see the emergence of more emotionally intelligent AI bots and new mental health applications driven by AI prediction and detection capabilities.

For instance, researchers at Vanderbilt University Medical Center in Tennessee, US, have developed an ML algorithm that uses a person's hospital admission data, including age, gender, and past medical diagnoses, to make an 80% accurate prediction of whether this individual is likely to take their own life. And researchers at the University of Florida are about to test their new AI platform aimed at making an accurate diagnosis in patients with early Parkinson's disease. Research is also underway to develop a tool combining explainable AI and deep learning to prescribe personalized treatment plans for children with schizophrenia.

AI systems could predict and plan treatments accurately and effectively for all fields of medicine at levels similar to that of physicians and general clinical practices. For example, one AI model demonstrated higher diagnostic accuracy for depression and post-traumatic stress disorder compared to general practitioners in controlled studies.

AI systems that analyze social media data are being developed to detect mental health risks more efficiently and cost-effectively across broader populations. Ethical concerns include uneven performance between digital services, the possibility that biases could affect decision-making, and trust, privacy, and doctor-patient relationship issues.

In January 2024, Cedars-Sinai physician-scientists developed a first-of-its-kind program that uses immersive virtual reality and generative AI to provide mental health support. The program is called XAIA which employs a large language model programmed to resemble a human therapist.

The University of Southern California has researched the effectiveness of a virtual therapist named Ellie. Through a webcam and microphone, this AI is able to process and analyze the emotional cues derived from the patient's face and the variation in expressions and tone of voice.

A team of Stanford psychologists and AI experts created "Woebot". Woebot is an app that makes therapy sessions available 24/7. WoeBot tracks its users' mood through brief daily chat conversations and offers curated videos or word games to assist users in managing their mental health. A Scandinavian team of software engineers and a clinical psychologist created "Heartfelt Services". Heartfelt Services is an application meant to simulate conventional talk therapy with an AI therapist.

Incorporating AI with EHR records, genomic data and clinical prescriptions can contribute to precision treatment. "Oura Ring", a wearable technology, scans the individual's heart rate and sleep routine in real time to give tailored suggestions. Such AI-based application has an increasing potential in combating the stigma of mental health.

Outcome comparisons: AI vs traditional therapy

Research shows that AI-driven mental health tools, particularly those using cognitive behavioral therapy (CBT), can improve symptoms of anxiety and depression, especially for mild to moderate cases. For example, chatbot-based interventions like Woebot significantly reduced depressive symptoms in young adults within two weeks, with results comparable to brief human-delivered interventions. A 2022 meta-analysis of digital mental health tools, including AI-enhanced apps, found moderate effectiveness in reducing symptoms when user engagement was high, and interventions were evidence-based.

However, traditional therapy remains more effective for complex or high-risk mental health conditions that require emotional nuance and relational depth, such as PTSD, severe depression, or suicidality. The therapeutic alliance, or the relationship between patient and clinician, is frequently cited in clinical literature as a significant factor in treatment outcomes, accounting for up to 30% of positive outcomes. While AI tools are capable of detecting patterns in behavior and speech, they are currently limited in replicating emotional nuance and the social context sensitivity typically provided by human clinicians. As such, most experts view AI in mental health as a complementary tool, best used for screening, monitoring, or augmenting care between human-led sessions.

While AI systems excel at processing large datasets and providing consistent, round-the-clock support, their rigidity and limitations in contextual understanding remain significant barriers. Human therapists can adapt in real time to tone, body language, and life circumstances—something machine learning models have yet to master. Nonetheless, integrated models that pair AI-driven symptom tracking with clinician oversight are showing promise. These hybrid approaches may increase access, reduce administrative burden, and support early detection, allowing human clinicians to focus on relational care. Current research suggests that AI in mental health care is more likely to augment rather than replace clinician-led therapy, particularly by supporting data analysis and continuous monitoring.

Criticism

Although artificial intelligence in mental health is a growing field with significant potential, several concerns and criticisms remain regarding its application:

• Data limitations: A significant barrier to developing effective AI tools in mental health care is the limited availability of high-quality, representative data. Mental health data is often sensitive, difficult to standardize, and subject to privacy restrictions, which can hinder the training of robust and generalizable AI models.
• Algorithmic bias: AI systems may inherit and amplify biases present in the datasets they are trained on. This can result in inaccurate assessments or unequal treatment, particularly for underrepresented or marginalized groups. It is important for developments in mental healthcare to be ethically valid. Major ethical concerns are breach of data privacy, bias in data algorithms, unlawful data access and stigma around mental health treatment. Algorithmic biases can result in misdiagnoses and incorrect treatment which are dangerous. One way to mitigate this is by ensuring that medical data is not segregated based on patient demographics. Another is to get rid of the binary gendering method and ensuring higher ups are informed of any developments in AI tech to avoid bias in the models. Creating a justified system where AI advances ethically, with its real-world applications helping instead of replacing medical professionals needs to be a priority.
• Privacy and data security: The implementation of AI in mental health typically requires the collection and analysis of large amounts of personal and sensitive information. This raises ethical concerns regarding user consent, data protection, and potential misuse of information.
• Risk of harmful advice: Some AI-based mental health tools have been criticized for offering inappropriate or harmful guidance. For example, there have been reports of chatbots giving users dangerous and even deadly recommendations, including one case in which a man died by suicide after a chatbot allegedly encouraged self-sacrifice, and multiple suicide cases in which ChatGPT reportedly encouraged victims to take their own lives, supplied victims with information on suicide methods, and/or urged victims to keep their suicidal ideations secret. In response to such incidents, several AI mental health applications have been taken offline or reevaluated for safety.
• Therapeutic relationship: Decades of psychological research have shown that the quality of the therapeutic relationship empathy, trust, and human connection is one of the most important predictors of treatment outcomes. Some researchers have questioned whether AI systems can replicate the relational dynamics shown to contribute to positive treatment outcomes. Medical professionals are expected to be empathetic and compassionate when interacting with their patients. However, certain authors have said that people interact with chatbots, fully aware that they are incapable of being genuinely empathetic like a human being and do not expect them to be sentient in their responses. Other authors have implied that it is illogical to expect patients to be emotionally vulnerable and open to chatbots. Only medical professionals have the human "touch" that helps them understand the "x factor" of their patients that machines cannot do. The possibility that therapists and medical professionals could be too emotionally exhausted at the end of the day to show their patients the compassion they are entitled to also exists. AI models and chatbots could have the advantage here. Maintaining a balance between the use of AI models and employing health professionals is important.
• Lack of emotional understanding: Unlike human therapists, AI systems do not possess lived experiences or emotional awareness that make them limited. These limitations have prompted debate about the role of AI in addressing emotionally complex mental health needs. Some experts argue that AI cannot substitute for human-centered therapy, particularly in cases requiring deep emotional engagement.
• Risk of psychosis: ChatGPT usage has driven some users to experience delusions. The realism of the interaction can leave a user believing that a real person is chatting with them, fueling cognitive dissonance. Some ChatGPT conversations endorsed conspiracies and mystical beliefs, and in some cases lead to suicide. Delusions and psychosis induced by AI usage has been referred to as chatbot psychosis.

Ethical issues

AI in mental health is progressing with personalized care to incorporate voice, speech and biometric data. But to prevent algorithmic bias, models need to be culturally inclusive too. Ethical issues, practical uses and bias in generative models need to be addressed to promote fair and reliable mental healthcare.

Although significant progress is still required, the integration of AI in mental health underscores the need for legal and regulatory frameworks to guide its development and implementation. Achieving a balance between human interaction and AI in healthcare is challenging, as there is a risk that increased automation may lead to a more mechanized approach, potentially diminishing the human touch that has traditionally characterized the field. Furthermore, granting patients a feeling of security and safety is a priority considering AI's reliance on individual data to perform and respond to inputs. Some experts caution that efforts to increase accessibility through automation may unintentionally affect aspects of the patient experience, such as trust or perceived support. To avoid veering in the wrong direction, more research should continue to develop a deeper understanding of where the incorporation of AI produces advantages and disadvantages.

Data privacy and confidentiality are one of the most common security threats to medical data. Chatbots are known to be used as virtual assistants for patients but the sensitive data they collect may not be protected because the US law does not consider them as medical devices. Pharmaceutical companies use this loophole to access sensitive information and use it for their own purpose which results, in a lack of trust in chatbots and patients can hesitate in providing information essential to their treatment. Conversational Artificial Intelligence stores and remembers every conversation with a patient with complete accuracy, smartphones also collect data from search history and track app activity. If such private information is leaked it could further increase the stigma around mental health. The danger of cybercrimes and the government's unprotected access to our data, all raise serious concerns about data security.

Additionally, a lack of clarity and openness with AI models can lead to a loss of trust from the patient for their medical advisors or doctors as the regular person is unaware of how they reach conclusions into giving certain medical advice. Access to such information is necessary to build trust. However, many of these models act like "black boxes", providing very little insight into how they work. AI specialists have thus highlighted ethical standards, diverse data and the correct usage of AI tools in mental healthcare.

Bias and discrimination

Artificial intelligence has shown promise in transforming mental health care through tools that support diagnosis, symptom tracking, and personalized interventions. However, significant concerns remain about the ways these systems may inadvertently reinforce existing disparities in care. Because AI models rely heavily on training data, they are particularly vulnerable to bias if that data fails to reflect the full range of racial, cultural, gender, and socioeconomic diversity found in the general population.

For example, a 2024 study from the University of California found that AI systems analyzing social media data to detect depression exhibited significantly reduced accuracy for Black Americans compared to white users, due to differences in language patterns and cultural expression that were not adequately represented in the training data. Similarly, natural language processing (NLP) models used in mental health settings may misinterpret dialects or culturally specific forms of communication, leading to misdiagnoses or missed signs of distress. These kinds of errors can compound existing disparities, particularly for marginalized populations that already face reduced access to mental health services.

Biases can also emerge during the design and deployment phases of AI development. Algorithms may inherit the implicit biases of their creators or reflect structural inequalities present in health systems and society at large. These issues have led to increased calls for fairness, transparency, and equity in the development of mental health technologies.

In response, researchers and healthcare institutions are taking steps to address bias and promote more equitable outcomes. Key strategies include:

• Inclusive data practices: Developers are working to curate and utilize datasets that reflect diverse populations in terms of race, ethnicity, gender identity, and socioeconomic background. This approach helps improve the generalizability and fairness of AI models.
• Bias assessment and auditing: Frameworks are being introduced to identify and mitigate algorithmic bias across the lifecycle of AI tools. This includes both internal validation (within training data) and external validation across new, diverse populations.
• Community and stakeholder engagement: Some projects now prioritize involving patients, clinicians, and representatives from underrepresented communities in the design, testing, and implementation phases. This helps ensure cultural relevance and supports greater trust in AI-assisted tools.
• Transparency and explainability: New efforts focus on building "explainable AI" systems that provide interpretable results and justifications for clinical decisions, allowing patients and providers to better understand and challenge AI-generated outcomes.

These efforts are still in early stages, but they reflect a growing recognition that equity must be a foundational principle in the deployment of AI in mental health care. When designed thoughtfully, AI systems could eventually help reduce disparities in care by identifying underserved populations, tailoring interventions, and increasing access in remote or marginalized communities. Continued investment in ethical design, oversight, and participatory development will be essential to ensure that AI tools do not replicate historical injustices but instead help move mental health care toward greater equity.

Entropic force

From Wikipedia, the free encyclopedia

https://en.wikipedia.org/wiki/Entropic_force 

In physics, an entropic force acting in a system is an emergent phenomenon resulting from the entire system's statistical tendency to increase its entropy, rather than from a particular underlying force on the atomic scale.

Mathematical formulation

In the canonical ensemble, the entropic force ${\displaystyle \mathbf{F}}$ associated to a macrostate partition ${\displaystyle \{\mathbf{X}\}}$ is given by

${\displaystyle \mathbf{F}({\mathbf{X}}_0)=T{\mathrm{\nabla }}_{\mathbf{X}}S(\mathbf{X}){|}_{{\mathbf{X}}_0},}$

where ${\displaystyle T}$ is the temperature, ${\displaystyle S(\mathbf{X})}$ is the entropy associated to the macrostate ${\displaystyle \mathbf{X}}$, and ${\displaystyle {\mathbf{X}}_0}$ is the present macrostate.

Examples

Pressure of an ideal gas

The internal energy of an ideal gas depends only on its temperature, and not on the volume of its containing box, so it is not an energy effect that tends to increase the volume of the box as gas pressure does. This implies that the pressure of an ideal gas has an entropic origin.

What is the origin of such an entropic force? The most general answer is that the effect of thermal fluctuations tends to bring a thermodynamic system toward a macroscopic state that corresponds to a maximum in the number of microscopic states (or micro-states) that are compatible with this macroscopic state. In other words, thermal fluctuations tend to bring a system toward its macroscopic state of maximum entropy.

Brownian motion

The entropic approach to Brownian movement was initially proposed by R. M. Neumann. Neumann derived the entropic force for a particle undergoing three-dimensional Brownian motion using the Boltzmann equation, denoting this force as a diffusional driving force or radial force. In the paper, three example systems are shown to exhibit such a force:

• electrostatic system of molten salt,
• surface tension and,
• elasticity of rubber.

Polymers

Main article: Ideal chain

A standard example of an entropic force is the elasticity of a freely jointed polymer molecule. For an ideal chain, maximizing its entropy means reducing the distance between its two free ends. Consequently, a force that tends to collapse the chain is exerted by the ideal chain between its two free ends. This entropic force is proportional to the distance between the two ends. The entropic force by a freely jointed chain has a clear mechanical origin and can be computed using constrained Lagrangian dynamics. With regards to biological polymers, there appears to be an intricate link between the entropic force and function. For example, disordered polypeptide segments – in the context of the folded regions of the same polypeptide chain – have been shown to generate an entropic force that has functional implications.

Hydrophobic force

See also: Hydrophobic effect § Cause

Water drops on the surface of grass

Another example of an entropic force is the hydrophobic force. At room temperature, it partly originates from the loss of entropy by the 3D network of water molecules when they interact with molecules of dissolved substance. Each water molecule is capable of

• donating two hydrogen bonds through the two protons,
• accepting two more hydrogen bonds through the two sp³-hybridized lone pairs.

Therefore, water molecules can form an extended three-dimensional network. Introduction of a non-hydrogen-bonding surface disrupts this network. The water molecules rearrange themselves around the surface, so as to minimize the number of disrupted hydrogen bonds. This is in contrast to hydrogen fluoride (which can accept 3 but donate only 1) or ammonia (which can donate 3 but accept only 1), which mainly form linear chains.

If the introduced surface had an ionic or polar nature, there would be water molecules standing upright on 1 (along the axis of an orbital for ionic bond) or 2 (along a resultant polarity axis) of the four sp³ orbitals. These orientations allow easy movement, i.e. degrees of freedom, and thus lowers entropy minimally. But a non-hydrogen-bonding surface with a moderate curvature forces the water molecule to sit tight on the surface, spreading 3 hydrogen bonds tangential to the surface, which then become locked in a clathrate-like basket shape. Water molecules involved in this clathrate-like basket around the non-hydrogen-bonding surface are constrained in their orientation. Thus, any event that would minimize such a surface is entropically favored. For example, when two such hydrophobic particles come very close, the clathrate-like baskets surrounding them merge. This releases some of the water molecules into the bulk of the water, leading to an increase in entropy.

Another related and counter-intuitive example of entropic force is protein folding, which is a spontaneous process and where hydrophobic effect also plays a role. Structures of water-soluble proteins typically have a core in which hydrophobic side chains are buried from water, which stabilizes the folded state. Charged and polar side chains are situated on the solvent-exposed surface where they interact with surrounding water molecules. Minimizing the number of hydrophobic side chains exposed to water is the principal driving force behind the folding process, although formation of hydrogen bonds within the protein also stabilizes protein structure.

Colloids

Entropic forces are important and widespread in the physics of colloids, where they are responsible for the depletion force, and the ordering of hard particles, such as the crystallization of hard spheres, the isotropic-nematic transition in liquid crystal phases of hard rods, and the ordering of hard polyhedra. Because of this, entropic forces can be an important driver of self-assembly

Entropic forces arise in colloidal systems due to the osmotic pressure that comes from particle crowding. This was first discovered in, and is most intuitive for, colloid-polymer mixtures described by the Asakura–Oosawa model. In this model, polymers are approximated as finite-sized spheres that can penetrate one another, but cannot penetrate the colloidal particles. The inability of the polymers to penetrate the colloids leads to a region around the colloids in which the polymer density is reduced. If the regions of reduced polymer density around two colloids overlap with one another, by means of the colloids approaching one another, the polymers in the system gain an additional free volume that is equal to the volume of the intersection of the reduced density regions. The additional free volume causes an increase in the entropy of the polymers, and drives them to form locally dense-packed aggregates. A similar effect occurs in sufficiently dense colloidal systems without polymers, where osmotic pressure also drives the local dense packing of colloids into a diverse array of structures that can be rationally designed by modifying the shape of the particles. These effects are for anisotropic particles referred to as directional entropic forces.

Cytoskeleton

Contractile forces in biological cells are typically driven by molecular motors associated with the cytoskeleton. However, a growing body of evidence shows that contractile forces may also be of entropic origin. The foundational example is the action of microtubule crosslinker Ase1, which localizes to microtubule overlaps in the mitotic spindle. Molecules of Ase1 are confined to the microtubule overlap, where they are free to diffuse one-dimensionally. Analogically to an ideal gas in a container, molecules of Ase1 generate pressure on the overlap ends. This pressure drives the overlap expansion, which results in the contractile sliding of the microtubules. An analogous example was found in the actin cytoskeleton. Here, the actin-bundling protein anillin drives actin contractility in cytokinetic rings.

Controversial examples

Some forces that are generally regarded as conventional forces have been argued to be actually entropic in nature. These theories remain controversial and are the subject of ongoing work. Matt Visser, professor of mathematics at Victoria University of Wellington, NZ in "Conservative Entropic Forces" criticizes selected approaches but generally concludes:

There is no reasonable doubt concerning the physical reality of entropic forces, and no reasonable doubt that classical (and semi-classical) general relativity is closely related to thermodynamics. Based on the work of Jacobson, Thanu Padmanabhan, and others, there are also good reasons to suspect a thermodynamic interpretation of the fully relativistic Einstein equations might be possible.

Gravity

Main article: Entropic gravity

In 2009, Erik Verlinde argued that gravity can be explained as an entropic force. It claimed (similar to Jacobson's result) that gravity is a consequence of the "information associated with the positions of material bodies". This model combines the thermodynamic approach to gravity with Gerard 't Hooft's holographic principle. It implies that gravity is not a fundamental interaction, but an emergent phenomenon.

Other forces

In the wake of the discussion started by Verlinde, entropic explanations for other fundamental forces have been suggested, including Coulomb's law. The same approach was argued to explain dark matter, dark energy and Pioneer effect.

Links to adaptive behavior

It was argued that causal entropic forces lead to spontaneous emergence of tool use and social cooperation. Causal entropic forces by definition maximize entropy production between the present and future time horizon, rather than just greedily maximizing instantaneous entropy production like typical entropic forces.

A formal simultaneous connection between the mathematical structure of the discovered laws of nature, intelligence and the entropy-like measures of complexity was previously noted in 2000 by Andrei Soklakov in the context of Occam's razor principle.

Evolution of human intelligence

From Wikipedia, the free encyclopedia

https://en.wikipedia.org/wiki/Evolution_of_human_intelligence 

The evolution of human intelligence is closely tied to the evolution of the human brain and to the origin of language. The timeline of human evolution spans approximately seven million years, from the separation of the genus Pan until the emergence of behavioral modernity by 50,000 years ago. The first three million years of this timeline concern Sahelanthropus, the following two million concern Australopithecus and the final two million span the history of the genus Homo in the Paleolithic era.

Many traits of human intelligence, such as empathy, theory of mind, mourning, ritual, and the use of symbols and tools, are somewhat apparent in other great apes, although they are observed in much less sophisticated forms than what is found in humans.

History

Hominidae

Hominin timeline
−10 — – −9 — – −8 — – −7 — – −6 — – −5 — – −4 — – −3 — – −2 — – −1 — – 0 —	Miocene Pliocene Pleistocene Hominini Nakalipithecus Samburupithecus Ouranopithecus (Ou. turkae) (Ou. macedoniensis) Chororapithecus Oreopithecus Sivapithecus Sahelanthropus Graecopithecus Orrorin (O. praegens) (O. tugenensis) Ardipithecus (Ar. kadabba) (Ar. ramidus) Australopithecus (Au. africanus) (Au. afarensis) (Au. anamensis) H. habilis (H. rudolfensis) (Au. garhi) H. erectus (H. antecessor) (H. ergaster) (Au. sediba) H. heidelbergensis Homo sapiens Neanderthals Denisovans	← Earlier apes ← Gorilla split ← Chimpanzee split ← Earliest bipedal ← Earliest sign of Ardipithecus ← Earliest sign of Australopithecus ← Earliest stone tools ← Earliest sign of Homo ← Dispersal beyond Africa ← Earliest fire / cooking ← Earliest rock art ← Earliest clothes ← Modern humans H o m i n i d s P a r a n t h r o p u s
(million years ago)

Further information: Primate empathy, Great ape language, and Lucy Temerlin

The great apes (Hominidae) show some cognitive and empathic abilities. Chimpanzees can make tools and use them to acquire foods and for social displays; they have mildly complex hunting strategies requiring cooperation, influence and rank; they are status conscious, manipulative and capable of deception; they can learn to use symbols and understand aspects of human language including some relational syntax, concepts of number and numerical sequence. One common characteristic that is present in species of "high degree intelligence" (i.e. dolphins, great apes, and humans - Homo sapiens) is a brain of enlarged size. Additionally, these species have a more developed neocortex, a folding of the cerebral cortex, and von Economo neurons. Said neurons are linked to social intelligence and the ability to gauge what another is thinking or feeling and are also present in bottlenose dolphins.

Homininae

Chimpanzee mother and baby

Around 10 million years ago, the Earth's climate entered a cooler and drier phase, which led eventually to the Quaternary glaciation beginning some 2.6 million years ago. One consequence of this was that the north African tropical forest began to retreat, being replaced first by open grasslands and eventually by desert (the modern Sahara). As their environment changed from continuous forest to patches of forest separated by expanses of grassland, some primates adapted to a partly or fully ground-dwelling life where they were exposed to predators, such as the big cats, from whom they had previously been safe.

These environmental pressures caused selection to favor bipedalism - walking on hind legs. This gave the Homininae's eyes greater elevation, the ability to see approaching danger further off, and a more efficient means of locomotion. It also freed their arms from the task of walking and made the hands available for tasks such as gathering food. At some point the bipedal primates developed handedness, giving them the ability to pick up sticks, bones and stones and use them as weapons, or as tools for tasks such as killing smaller animals, cracking nuts, or cutting up carcasses. In other words, these primates developed the use of primitive technology. Bipedal tool-using primates from the subtribe Hominina date back to as far as about 5 to 7 million years ago, such as one of the earliest species, Sahelanthropus tchadensis.

From about 5 million years ago, the hominin brain began to develop rapidly in both size and differentiation of function. There has been a gradual increase in brain volume as humans progressed along the timeline of evolution (see Homininae), starting from about 600 cm³ in Homo habilis up to 1500 cm³ in Homo neanderthalensis. Thus, in general there's a positive correlation between brain volume and intelligence. However, modern Homo sapiens have a brain volume slightly smaller (1250 cm³) than neanderthals, and the Flores hominids (Homo floresiensis), nicknamed hobbits, had a cranial capacity of about 380 cm³ (considered small for a chimpanzee) about a third of that of Homo erectus. It is proposed that they evolved from H. erectus as a case of insular dwarfism. With their three-times-smaller brain, the Flores hominids apparently used fire and made tools as sophisticated as those of their ancestor H. erectus.

Homo

Further information: Homo

Roughly 2.4 million years ago Homo habilis had appeared in East Africa: the first known human species, and the first known to make stone tools, yet the disputed findings of signs of tool use from even earlier ages and from the same vicinity as multiple Australopithecus fossils may put to question how much more intelligent than its predecessors H. habilis was.

The use of tools conferred a crucial evolutionary advantage, and required a larger and more sophisticated brain to co-ordinate the fine hand movements required for this task. Our knowledge of the complexity of behaviour of Homo habilis is not limited to stone culture; they also had habitual therapeutic use of toothpicks.

A larger brain requires a larger skull, and thus is accompanied by other morphological and biological evolutionary changes. One such change required for the female to have a wider birth canal for the newborn's larger skull to pass through. The solution to this was to give birth at an early stage of fetal development, before the skull grew too large to pass through the birth canal. Other accompanying adaptations were the smaller maxillary and mandibular bones, smaller and weaker facial muscles, and shortening and flattening of the face resulting in modern-human's complex cognitive and linguistic capabilities as well as the ability to create facial expressions and smile. Consequentially, dental issues in modern humans arise from these morphological changes that are exacerbated by a shift from nomadic to sedentary lifestyles.

Humans' increasingly sedentary lifestyle to protect their more vulnerable offspring led them to grow even more dependent on tool-making to compete with other animals and other humans, and rely less on body size and strength.

About 200,000 years ago Europe and the Middle East were colonized by Neanderthals, extinct by 39,000 years ago following the appearance of modern humans in the region from 40,000 to 45,000 years ago.

History of humans

In the Late Pliocene, hominins were set apart from modern great apes and other closely related organisms by the anatomical evolutionary changes resulting in bipedalism, or the ability to walk upright. Characteristics such as a supraorbital torus, or prominent eyebrow ridge, and flat face also makes Homo erectus distinguishable. Their brain size substantially sets them apart from closely related species, such as H. habilis, as seen by an increase in average cranial capacity of 1000 cc. Compared to earlier species, H. erectus developed keels and small crests in the skull showing morphological changes of the skull to support increased brain capacity. It is believed that Homo erectus were, anatomically, modern humans as they are very similar in size, weight, bone structure, and nutritional habits. Over time, however, human intelligence developed in phases that is interrelated with brain physiology, cranial anatomy and morphology, and rapidly changing climate and environments.

Drawing of Acheulean handaxe from Spain from front, back, side, and top profile

Tool-use

The study of the evolution of cognition relies on the archaeological record made up of assemblages of material culture, particularly from the Paleolithic Period, to make inferences about our ancestors' cognition. Paleo-anthropologists from the past half-century have had the tendency of reducing stone tool artifacts to physical products of the metaphysical activity taking place in the brains of hominins. Recently, a new approach called 4E cognition (see Models for other approaches) has been developed by cognitive archaeologists Lambros Malafouris, Thomas G. Wynn, and Karenleigh A. Overmann, to move past the "internal" and "external" dichotomy by treating stone tools as objects with agency in both providing insight to hominin cognition and having a role in the development of early hominin cognition. The 4E cognition approach describes cognition as embodied, embedded, enactive, and extended, to understand the interconnected nature between the mind, body, and environment.

There are four major categories of tools created and used throughout human evolution that are associated with the corresponding evolution of the brain and intelligence. Stone tools such as flakes and cores used by Homo habilis for cracking bones to extract marrow, known as the Oldowan culture, make up the oldest major category of tools from about 2.5 and 1.6 million years ago. The development of stone tool technology suggests that our ancestors had the ability to hit cores with precision, taking into account the force and angle of the strike, and the cognitive planning and capacity to envision a desired outcome.

Stone tools from the Paleolithic Period, also known as the Stone Age, are indicative of cognitive advancements throughout human evolutionary history.

Acheulean culture, associated with Homo erectus, is composed of bifacial, or double-sided, hand-axes, that "requires more planning and skill on the part of the toolmaker; he or she would need to be aware of principles of symmetry". In addition, some sites show evidence that selection of raw materials involved travel, advanced planning, cooperation, and thus communication with other hominins.

The third major category of tool industry marked by its innovation in tool-making technique and use is the Mousterian culture. Compared to previous tool cultures, in which tools were regularly discarded after use, Mousterian tools, associated with Neanderthals, were specialized, built to last, and "formed a true toolkit". The making of these tools, called the Levallois technique, involves a multi-step process which yields several tools. In combination with other data, the formation of this tool culture for hunting large mammals in groups evidences the development of speech for communication and complex planning capabilities.

While previous tool cultures did not show great variation, the tools of early modern Homo sapiens are robust in the amount of artifacts and diversity in utility. There are several styles associated with this category of the Upper Paleolithic, such as blades, boomerangs, atlatls (spear throwers), and archery made from varying materials of stone, bone, teeth, and shell. Beyond use, some tools have been shown to have served as signifiers of status and group membership. The role of tools for social uses signal cognitive advancements such as complex language and abstract relations to things.

Homo sapiens

"The Lion-man", found in the Hohlenstein-Stadel cave of Germany's Swabian Alb and dated to 40,000 years ago, is associated with the Aurignacian culture and is the oldest known anthropomorphic animal figurine in the world.

Dates approximate, consult articles for details

(From 2,000,000 BC to 2013 AD in (partial) exponential notation)

See also: Java Man (−1.75e+06), Yuanmou Man (−1.75e+06: -0.73e+06),

Lantian Man (−1.7e+06), Nanjing Man (- 0.6e+06), Tautavel Man (- 0.5e+06),

Peking Man (- 0.4e+06), Solo Man (- 0.4e+06), and Peștera cu Oase (- 0.378e+05)

Homo sapiens intelligence

Further information: Intelligence, Archaic humans, Behavioral modernity, and Early human migration

The eldest findings of Homo sapiens in Jebel Irhoud, Morocco date back c. 300,000 years. Fossils of Homo sapiens were found in East Africa which are c. 200,000 years old. It is unclear to what extent these early modern humans had developed language, music, religion, etc. The cognitive tradeoff hypothesis proposes that there was an evolutionary tradeoff between short-term working memory and complex language skills over the course of human evolution.

According to proponents of the Toba catastrophe theory, the climate in non-tropical regions of the earth experienced a sudden freezing about 70,000 years ago, because of a huge explosion of the Toba volcano that filled the atmosphere with volcanic ash for several years. This reduced the human population to less than 10,000 breeding pairs in equatorial Africa, from which all modern humans are descended. Being unprepared for the sudden change in climate, the survivors were those intelligent enough to invent new tools and ways of keeping warm and finding new sources of food (for example, adapting to ocean fishing based on prior fishing skills used in lakes and streams that became frozen).

Around 80,000–100,000 years ago, three main lines of Homo sapiens diverged, bearers of mitochondrial haplogroup L1 (mtDNA) / A (Y-DNA) colonizing Southern Africa (the ancestors of the Khoisan/Capoid peoples), bearers of haplogroup L2 (mtDNA) / B (Y-DNA) settling Central and West Africa (the ancestors of Niger–Congo and Nilo-Saharan speaking peoples), while the bearers of haplogroup L3 remained in East Africa.

The "Great Leap Forward" leading to full behavioral modernity sets in only after this separation. Rapidly increasing sophistication in tool-making and behaviour is apparent from about 80,000 years ago, and the migration out of Africa follows towards the very end of the Middle Paleolithic, some 60,000 years ago. Fully modern behaviour, including figurative art, music, self-ornamentation, trade, burial rites etc. is evident by 30,000 years ago. The oldest unequivocal examples of prehistoric art date to this period, the Aurignacian and the Gravettian periods of prehistoric Europe, such as the Venus figurines and cave painting (Chauvet Cave) and the earliest musical instruments (the bone pipe of Geissenklösterle, Germany, dated to about 36,000 years ago).

Motor and sensory areas of the cerebral cortex; dashed areas shown are commonly left hemisphere dominant.

The human brain has evolved gradually over the passage of time; a series of incremental changes occurring as a result of external stimuli and conditions. It is crucial to keep in mind that evolution operates within a limited framework at a given point in time. In other words, the adaptations that a species can develop are not infinite and are defined by what has already taken place in the evolutionary timeline of a species. Given the immense anatomical and structural complexity of the brain, its evolution (and the congruent evolution of human intelligence), can only be reorganized in a finite number of ways. The majority of said changes occur either in terms of size or in terms of developmental timeframes.

The cerebral cortex is divided into four lobes (frontal, parietal, occipital, and temporal) each with specific functions. The cerebral cortex is significantly larger in humans than in any other animal and is responsible for higher thought processes such as reasoning, abstract thinking, and decision making. Another characteristic that makes humans special and sets them apart from any other species is our ability to produce and understand complex, syntactic language. The cerebral cortex, particularly in the temporal, parietal, and frontal lobes, are populated with neural circuits dedicated to language. There are two main areas of the brain commonly associated with language, namely: Wernicke's area and Broca's area. The former is responsible for the understanding of speech and the latter for the production of speech. Homologous regions have been found in other species (i.e. Area 44 and 45 have been studied in chimpanzees) but they are not as strongly related to or involved in linguistic activities as in humans.

Models

Massive modularity of mind

Each card has a number on one side, and a patch of color on the other. Which card or cards must be turned over to test the idea that if a card shows an even number on one face, then its opposite face is blue?

 

Each card has an age on one side, and a drink on the other. Which card or cards must be turned over to test the idea that if someone is drinking alcohol then they must be over 18?

In 2004, psychologist Satoshi Kanazawa argued that g was a domain-specific, species-typical, information processing psychological adaptation, and in 2010, Kanazawa argued that g correlated only with performance on evolutionarily unfamiliar rather than evolutionarily familiar problems, proposing what he termed the "Savanna-IQ interaction hypothesis". In 2006, Psychological Review published a comment reviewing Kanazawa's 2004 article by psychologists Denny Borsboom and Conor Dolan that argued that Kanazawa's conception of g was empirically unsupported and purely hypothetical and that an evolutionary account of g must address it as a source of individual differences. In response to Kanazawa's 2010 article, psychologists Scott Barry Kaufman, Colin G. DeYoung, Deirdre Reis, and Jeremy R. Gray gave 112 subjects a 70-item computerized version of the Wason selection task (a logic puzzle) in a social relations context as proposed by Leda Cosmides and John Tooby in The Adapted Mind, and found instead that "performance on non-arbitrary, evolutionarily familiar problems is more strongly related to general intelligence than performance on arbitrary, evolutionarily novel problems".

Peter Cathcart Wason originally demonstrated that not even 10% of subjects found the correct solution and his finding was replicated. Psychologists Patricia Cheng, Keith Holyoak, Richard E. Nisbett, and Lindsay M. Oliver demonstrated experimentally that subjects who have completed semester-long college courses in propositional calculus do not perform better on the Wason selection task than subjects who do not complete such college courses. Tooby and Cosmides originally proposed a social relations context for the Wason selection task as part of a larger computational theory of social exchange after they began reviewing the previous experiments about the task beginning in 1983. Despite other experimenters finding that some contexts elicited more correct subject responses than others, no theoretical explanation for differentiating between them was identified until Tooby and Cosmides proposed that disparities in subjects performance on contextualized versus non-contextualized variations of the task was an artifact of the task measuring a specialized cheater-detection module. Tooby and Cosmides later noted that whether there are evolved cognitive mechanisms for the content-blind rules of logical inference is disputed, and consistently noted that a body of research about the Wason selection task had concluded that cognitive adaptations for social exchange were not a by-product of general-purpose reasoning mechanisms, domain-general learning mechanisms, or g.

Relatedly, economist Thomas Sowell has noted that numerous studies finding disparities between the mean test scores of ethnic groups on intelligence tests have found that ethnic groups with lower mean test scores have tended to perform worst on spatial, non-verbal, or abstract reasoning test items. Writing after the completion of the Human Genome Project in 2003, psychologist Earl B. Hunt noted in 2011 that no genes related to differences in cognitive skills across various racial and ethnic groups had ever been discovered. In 2012, American Psychologist published a review by Nisbett, psychologists Joshua Aronson, Clancy Blair, Diane F. Halpern, and Eric Turkheimer, economist William Dickens, and philosopher James R. Flynn of findings since the publication of the 1995 American Psychological Association report on intelligence that concluded that almost no single-nucleotide genetic polymorphisms that have been discovered are consistently associated with variation in IQ in the normal range, and that adoption research on race and intelligence showed that differences could be entirely accounted for by environmental factors. In 2021, subsequent research using polygenic scores for educational attainment and cognitive performance in African and European samples from the 1000 Genomes Project found no evidence of divergent selection by race and a statistically insignificant contribution to racial differences in IQ.

Flynn had argued earlier that the Flynn effect presented multiple paradoxes for g as a psychological trait with a heritable basis because the increases in the statistical average scores among later birth year cohorts born in the 20th century were occurring without sufficient increases in vocabulary size, general knowledge, and ability to solve arithmetical problems, and that the increases were so large that they would imply that the statistically average members of the birth year cohorts in the late 19th century and early 20th century (the Lost Generation and the Greatest Generation) would have been intellectually disabled (as well as more distant human ancestors). Hunt noted that the latter paradox would imply that half of the soldiers who served in the U.S. military during World War II would not pass the Armed Services Vocational Aptitude Battery in 2008. Flynn proposed that these paradoxes could be answered by the increasing use of abstraction, logic, and scientific reasoning to address problems, while Nisbett argued that the Flynn effect was largely attributable to increases in formal education among human populations during the 20th century.

In 2010, psychologist David Marks found through 8 statistical analyses that average population IQ scores across race, time, and nationality correlated with literacy rates between a range of 0.79 and 0.99, which led to the conclusion that both the Flynn effect and racial differences in mean scores on intelligence tests were statistical artifacts of uncontrolled variation in literacy rates due to test performance requiring literacy. However, in reference to theoretical issues with constructivism in mathematics education and the failure of whole language in literacy education, psychologist David C. Geary and cognitive scientist Steven Pinker have noted that literacy, numeracy, and formal mathematical and logical reasoning are not inherited traits but biologically secondary cognitive skills (i.e. acquired characteristics) that require extensive practice after formal, explicit, and direct instruction—in contrast with natural language and number sense, since language acquisition and numerosity develop automatically and unconsciously due to specialized neurobiological systems for language and numerical cognition which the biologically secondary cognitive skills lack. In research on Broca's area, Pinker and coauthors reconfirmed the findings about the visual word form area (VWFA) of cognitive neuroscientist Stanislas Dehaene, who proposed the neuronal recycling hypothesis as an explanation for the VWFA and the associations between the parietal lobe and intraparietal sulcus in mental arithmetic with numeral systems. Pinker and Geary also reference Dehaene's research on the development of arithmetical skills, dyscalculia, and acalculia.

Pinker has also noted that writing is not a cultural universal since writing systems were independently invented only a few times in human history and most societies documented by ethnographers lacked writing systems, while literacy rates in European countries did not begin to exceed 50 percent until the 17th century since the movable-type printing press was not invented until the 15th century. Similarly to the lack of improvement in performance on the Wason selection task by college students that take courses in propositional calculus, Pinker referenced the response by professional mathematicians and statisticians to the solution to the Monty Hall problem published in Parade in 1990 in noting the dominance of automatic processes over controlled processes for formal logical reasoning following the dual process model proposed by psychologists Daniel Kahneman and Amos Tversky. While Pinker has suggested that the evolution of human intelligence could be explained by intelligence itself being the product of metaphor (stemming from the ability to create arbitrary morphemes) and combinatorial grammar (allowing nesting of verb phrases in syntax) that together enable the infinite composition of sentences, Pinker has also argued that the Flynn effect is likely caused by increased amounts of formal education in addition to other factors.

Noting that Kaufman and psychologist Robert Sternberg identified a lack of consensus about how to define human intelligence, psychologists Jay Joseph and Ken Richardson have argued that the construct validity of intelligence tests is questionable due to definitions of intelligence being based on the intuitions of psychologists, that tests measure formal education more than innate intelligence because test items are included because of how such items reflect academic performance, and that the Flynn effect provides evidence against heritability causing even within-group differences in general intelligence due to the confounding effect of shared birth dates and nationalities between pairs of identical twins reared apart from different generations in test performance. While Hunt defended the methodology of twin and adoption research in behavioral genetics, Hunt also noted that the only candidate genes behavioral geneticists had proposed being related to g were associated with below average IQ scores and that none had been identified that were associated with above average IQ scores.

Although Hunt expressed some reservations about the construct validity of g (referencing the research of psychologist Louis Leon Thurstone) and acknowledged the impact of literacy on IQ test performance, Hunt defended the fluid and crystallized intelligence models of g in psychometrics, and argued that alternatives to psychometric models (such as the theory of multiple intelligences and the triarchic theory of intelligence) lacked empirical support. Hunt also argued that research on the evolution of the brain showed evidence for g as a general problem-solving mechanism. Conversely, Pinker has argued that research in cognitive neuroscience has shown that the brain is more characterized by functional specialization. While Geary has attempted to integrate g with evolutionary psychology, Tooby, Cosmides, and Pinker have all argued that the human mind is better understood as a system of dedicated intelligences and domain-specific learning systems that are adaptively specialized rather than characterized by a general intelligence factor and a domain-general learning system that enables the passive cultural learning and socialization of a blank slate.

Tooby and Cosmides suggested that the human mind does have a domain-general, content-independent, and general-purpose improvisational intelligence that resembles general intelligence, and which possibly evolved to generate solutions in novel situations where the dedicated intelligences did not produce an optimal response. However, in light of the frame problem and combinatorial explosion in artificial intelligence and because all adaptations require selection pressure from recurrent problems, Tooby and Cosmides argue that a complete blank slate mind entirely shaped by general intelligence following the standard social science model is not computationally capable of performing the cognitive tasks or solving the adaptive problems that the human mind evolved to perform and solve, such as visual perception, language acquisition, recognizing emotional expressions, mate choice, cultural learning, and cheater-detection in social exchange.

Social brain hypothesis

Further information: Dunbar's number

The social brain hypothesis was proposed by British anthropologist Robin Dunbar, who argues that human intelligence did not evolve primarily as a means to solve ecological problems, but rather as a means of surviving and reproducing in large and complex social groups. Some of the behaviors associated with living in large groups include reciprocal altruism, deception, and coalition formation. These group dynamics relate to Theory of Mind or the ability to understand the thoughts and emotions of others, though Dunbar himself admits in the same book that it is not the flocking itself that causes intelligence to evolve (as shown by ruminants).

Dunbar argues that when the size of a social group increases, the number of different relationships in the group may increase by orders of magnitude. Chimpanzees live in groups of about 50 individuals whereas humans typically have a social circle of about 150 people, which is also the typical size of social communities in small societies and personal social networks; this number is now referred to as Dunbar's number. In addition, there is evidence to suggest that the success of groups is dependent on their size at foundation, with groupings of around 150 being particularly successful, potentially reflecting the fact that communities of this size strike a balance between the minimum size of effective functionality and the maximum size for creating a sense of commitment to the community. According to the social brain hypothesis, when hominids started living in large groups, selection favored greater intelligence. As evidence, Dunbar cites a relationship between neocortex size and group size of various mammals.

Criticism

Phylogenetic studies of brain sizes in primates show that while diet predicts primate brain size, sociality does not predict brain size when corrections are made for cases in which diet affects both brain size and sociality. The exceptions to the predictions of the social intelligence hypothesis, which that hypothesis has no predictive model for, are successfully predicted by diets that are either nutritious but scarce or abundant but poor in nutrients. Researchers have found that frugivores tend to exhibit larger brain size than folivores. One potential explanation for this finding is that frugivory requires "extractive foraging", or the process of locating and preparing hard-shelled foods, such as nuts, insects, and fruit. Extractive foraging requires higher cognitive processing, which could help explain larger brain size. However, other researchers argue that extractive foraging was not a catalyst in the evolution of primate brain size, demonstrating that some non primates exhibit advanced foraging techniques. Other explanations for the positive correlation between brain size and frugivory highlight how the high-energy, frugivore diet facilitates fetal brain growth and requires spatial mapping to locate the embedded foods.

Meerkats have far more social relationships than their small brain capacity would suggest. Another hypothesis is that it is actually intelligence that causes social relationships to become more complex, because intelligent individuals are more difficult to learn to know.

There are also studies that show that Dunbar's number is not the upper limit of the number of social relationships in humans either.

The hypothesis that it is brain capacity that sets the upper limit for the number of social relationships is also contradicted by computer simulations that show simple unintelligent reactions to be sufficient to emulate "ape politics" and by the fact that some social insects such as the paper wasp do have hierarchies in which each individual has its place (as opposed to herding without social structure) and maintains their hierarchies in groups of approximately 80 individuals with their brains smaller than that of any mammal.

Insects provide an opportunity to explore this since they exhibit an unparalleled diversity of social forms to permanent colonies containing many individuals working together as a collective organism and have evolved an impressive range of cognitive skills despite their small nervous systems. Social insects are shaped by ecology, including their social environment. Studies aimed to correlating brain volume to complexity have failed to identify clear correlations between sociality and cognition because of cases like social insects. In humans, societies are usually held together by the ability of individuals to recognize features indicating group membership. Social insects, likewise, often recognize members of their colony allowing them to defend against competitors. Ants do this by comparing odors which require fine discrimination of multicomponent variable cues. Studies suggest this recognition is achieved through simple cognitive operations that do not involve long-term memory but through sensory adaptation or habituation. In honeybees, their symbolic 'dance' is a form of communication that they use to convey information with the rest of their colony. In an even more impressive social use of their dance language, bees indicate suitable nest locations to a swarm in search of a new home. The swarm builds a consensus from multiple 'opinions' expressed by scouts with different information, to finally agree on a single destination to which the swarm relocates.

Cultural intelligence hypothesis

Overview

See also: Orthogenesis

Similar to, but distinct from the social brain hypothesis, is the cultural intelligence or cultural brain hypothesis, which dictates that human brain size, cognitive ability, and intelligence have increased over generations due to cultural information from a mechanism known as social learning. The hypothesis also predicts a positive correlation between species with a higher dependency and more frequent opportunities for social learning and overall cognitive ability. This is because social learning allows species to develop cultural skills and strategies for survival; in this way it can be said that heavily cultural species should in theory be more intelligent.

Humans have been widely acknowledged as the most intelligent species on the planet, with big brains with ample cognitive abilities and processing power which outcompete all other species. In fact, humans have shown an enormous increase in brain size and intelligence over millions of years of evolution. This is because humans have been referred to as an 'evolved cultural species'; one that has an unrivalled reliance on culturally transmitted knowledge due to the social environment around us. This is down to social transmission of information which spreads significantly faster in human populations relative to changes in genetics. Put simply, humans are the most cultural species there is, and are therefore the most intelligent species there is. The key point when concerning evolution of intelligence is that this cultural information has been consistently transmitted across generations to build vast amounts of cultural skills and knowledge throughout the human race. Dunbar's social brain hypothesis on the other hand dictates that our brains evolved primarily due to complex social interactions in groups, so in this way the two hypotheses are distinct from each other in that the cultural intelligence hypothesis focuses more on an in increase in intelligence from socially transmitted information. A shift in focus from 'social' interactions to learning strategies can be seen through this. The hypothesis can also be seen to contradict the idea of human 'general intelligence' by emphasising the process of cultural skills and information being learned from others.

In 2018, Muthukrishna and researchers constructed a model based on the cultural intelligence hypothesis which revealed relationships between brain size, group size, social learning and mating structures. The model had three underlying assumptions:

1. Brain size, complexity and organisation were grouped into one variable
2. A larger brain results in larger capacity for adaptive knowledge
3. More adaptive knowledge increases fitness of organisms

Using evolutionary simulation, the researchers were able to confirm the existence of hypothesised relationships. Results concerning the cultural intelligence hypothesis model showed that larger brains can store more information and adaptive knowledge, thus supporting larger groups. This abundance of adaptive knowledge can then be used for frequent social learning opportunities.

Further empirical evidence

As previously mentioned, social learning is the foundation of the cultural intelligence hypothesis and can be described simplistically as learning from others. It involves behaviours such as imitation, observational learning, influences from family and friends and explicit teaching from others. What sets humans apart from other species is that, due to our emphasis on culturally acquired information, humans have evolved to already possess significant social learning abilities from infancy. Neurological studies on nine month old infants were conducted by researchers in 2012 to demonstrate this phenomenon. The study involved infants observing a caregiver making a sound with a rattle over a period of one week. The brains of the infants were monitored throughout the study. Researchers found that the infants were able to activate neural pathways associated with making a sound with the rattle without actually doing the action themselves, showing human social learning in action- infants were able to understand the effects of a particular action simply by observing the performance of the action by someone else. Not only does this study demonstrate the neural mechanisms of social learning, but it also demonstrates our inherent ability to acquire cultural skills from those around us from the very start of our lives- it therefore shows strong support for the cultural intelligent hypothesis.

Various studies have been conducted to show the cultural intelligence hypothesis in action on a wider scale. One particular study in 2016 investigated two orangutan species, including the more social Sumatran species and the less sociable Bornean species. The aim was to test the notion that species with a higher frequency of opportunities for social learning should evolve to be more intelligent. Results showed that the Sumatrans consistently performed better in cognitive tests compared to the less sociable Borneans. The Sumatrans also showed greater inhibition and more cautious behaviour within their habitat. This was one of the first studies to show evidence for the cultural intelligence hypothesis in a non human species- frequency of learning opportunities had gradually produced differences in cognitive abilities between the two species.

Transformative cultural intelligence hypothesis

A study in 2018 proposed an altered variant of the original version of the hypothesis called the 'transformative cultural intelligence hypothesis'. The research involved investigating four year old's problem solving skills in different social contexts. The children were asked to extract a floating object from a tube using water. Nearly all were unsuccessful without cues, however most children succeeded after being shown a pedagogical solution suggesting video. When the same video was shown in a non pedagogical manner however, the children's success in the task did not improve. Crucially, this meant that the children's physical cognition and problem solving ability was therefore affected by how the task was socially presented to them. Researchers thus formulated the transformative cultural intelligence hypothesis, which stresses that our physical cognition is developed and affected by the social environment around us. This challenges the traditional cultural intelligence hypothesis which states that it is human's social cognition and not physical cognition which is superior to our nearest primate relatives; showing unique physical cognition in humans affected by external social factors. This phenomenon has not been seen in other species.

Reduction in aggression

Another theory that tries to explain the growth of human intelligence is the reduced aggression theory (aka self-domestication theory). According to this strand of thought, what led to the evolution of advanced intelligence in Homo sapiens was a drastic reduction of the aggressive drive. This change separated us from other species of monkeys and primates, where this aggressivity is still in plain sight, and eventually lead to the development of quintessential human traits such as empathy, social cognition, and culture. This theory has received strong support from studies of animal domestication where selective breeding for tameness has, in only a few generations, led to the emergence of impressive "humanlike" abilities. Tamed foxes, for example, exhibit advanced forms of social communication (following pointing gestures), pedomorphic physical features (childlike faces, floppy ears) and even rudimentary forms of theory of mind (eye contact seeking, gaze following). Evidence also comes from the field of ethology (which is the study of animal behavior, focused on observing species in their natural habitat rather than in controlled laboratory settings) where it has been found that animals with a gentle and relaxed manner of interacting with each other – for example stumptailed macaques, orangutans and bonobos – have more advanced socio-cognitive abilities than those found among the more aggressive chimpanzees and baboons. It is hypothesized that these abilities derive from a selection against aggression.

On a mechanistic level, these changes are believed to be the result of a systemic downregulation of the sympathetic nervous system (the fight-or-flight reflex). Hence, tamed foxes show a reduced adrenal gland size and have an up to fivefold reduction in both basal and stress-induced blood cortisol levels. Similarly, domesticated rats and guinea pigs have both reduced adrenal gland size and reduced blood corticosterone levels. It seems as though the neoteny of domesticated animals significantly prolongs the immaturity of their hypothalamic-pituitary-adrenal system (which is otherwise only immature for a short period when they are pups/kittens) and this opens up a larger "socialization window" during which they can learn to interact with their caretakers in a more relaxed way.

This downregulation of sympathetic nervous system reactivity is also believed to be accompanied by a compensatory increase in a number of opposing organs and systems. Although these are not as well specified, various candidates for such "organs" have been proposed: the parasympathetic system as a whole, the septal area over the amygdala, the oxytocin system, the endogenous opioids and various forms of quiescent immobilization which antagonize the fight-or-flight reflex.

Sexual selection

Main articles: Sexual selection in humans and Human mating strategies

This model, which invokes sexual selection, is proposed by Geoffrey Miller who argues that human intelligence is unnecessarily sophisticated for the needs of hunter-gatherers to survive. He argues that the manifestations of intelligence such as language, music and art did not evolve because of their utilitarian value to the survival of ancient hominids. Rather, intelligence may have been a fitness indicator. Hominids would have been chosen for greater intelligence as an indicator of healthy genes and a Fisherian runaway positive feedback loop of sexual selection would have led to the evolution of human intelligence in a relatively short period. Philosopher Denis Dutton also argued that the human capacity for aesthetics evolved by sexual selection.

Evolutionary biologist George C. Williams and evolutionary psychiatrist Randolph M. Nesse cite evolutionary psychologists John Tooby and Leda Cosmides as referring to the emotions as "Darwinian algorithms of the mind", while social psychologist David Buss has argued that the sex-specialized differences in the emotion of jealousy are adaptive strategies for detecting infidelity by a mating partner and anthropologists Donald E. Brown and Ward Goodenough have argued that marriage is a cultural universal that evolved to regulate sexual access to fertile women within a particular culture in response to male intrasexual competition and dominance. Citing cross-cultural research conducted by Buss, Miller has argued that if humans prefer altruistic mating partners that would select by mate choice for altruism directly. Additionally, Nesse and theoretical biologist Mary Jane West-Eberhard view sexual selection as a subcategory of social selection,^[146] with Nesse and anthropologist Christopher Boehm arguing further that altruism in humans held fitness advantages that enabled evolutionarily extraordinary cooperativeness and the human capability of creating culture, as well as capital punishment by band societies against bullies, thieves, free-riders, and psychopaths.

In many species, only males have impressive secondary sexual characteristics such as ornaments and show-off behavior, but sexual selection is also thought to be able to act on females as well in at least partially monogamous species. With complete monogamy, there is assortative mating for sexually selected traits. This means that less attractive individuals will find other less attractive individuals to mate with. If attractive traits are good fitness indicators, this means that sexual selection increases the genetic load of the offspring of unattractive individuals. Without sexual selection, an unattractive individual might find a superior mate with few deleterious mutations, and have healthy children that are likely to survive. With sexual selection, an unattractive individual is more likely to have access only to an inferior mate who is likely to pass on many deleterious mutations to their joint offspring, who are then less likely to survive.

Sexual selection is often thought to be a likely explanation for other female-specific human traits, for example breasts and buttocks far larger in proportion to total body size than those found in related species of ape. It is often assumed that if breasts and buttocks of such large size were necessary for functions such as suckling infants, they would be found in other species. That human female breasts (typical mammalian breast tissue is small) are found sexually attractive by many men is in agreement with sexual selection acting on human females secondary sexual characteristics.

Sexual selection for intelligence and judging ability can act on indicators of success, such as highly visible displays of wealth. Growing human brains require more nutrition than brains of related species of ape. It is possible that for females to successfully judge male intelligence, they must be intelligent themselves. This could explain why despite the absence of clear differences in intelligence between males and females on average, there are clear differences between male and female propensities to display their intelligence in ostentatious forms.

Critique

The sexual selection by the disability principle/fitness display model of the evolution of human intelligence is criticized by certain researchers for issues of timing of the costs relative to reproductive age. While sexually selected ornaments such as peacock feathers and moose antlers develop either during or after puberty, timing their costs to a sexually mature age, human brains expend large amounts of nutrients building myelin and other brain mechanisms for efficient communication between the neurons early in life. These costs early in life build facilitators that reduce the cost of neuron firing later in life, and as a result the peaks of the brain's costs and the peak of the brain's performance are timed on opposite sides of puberty with the costs peaking at a sexually immature age while performance peaks at a sexually mature age. Critical researchers argue the above shows that the cost of intelligence is a signal which reduces the chance of surviving to reproductive age, and does not signal fitness of sexually mature individuals. Since the disability principle is about selection from disabilities in sexually immature individuals, which increases the offspring's chance of survival to reproductive age, disabilities would be selected against and not for by the above mechanism. These critics argue that human intelligence evolved by natural selection citing that unlike sexual selection, natural selection have produced many traits that cost the most nutrients before puberty including immune systems and accumulation and modification for increased toxicity of poisons in the body as a protective measure against predators.

Intelligence as a disease-resistance sign

The number of people with severe cognitive impairment caused by childhood viral infections like meningitis, protists like Toxoplasma and Plasmodium, and animal parasites like intestinal worms and schistosomes is estimated to be in the hundreds of millions. Even more people with moderate mental damages, such as an inability to complete difficult tasks, that are not classified as 'diseases' by medical standards, may still be considered as inferior mates by potential sexual partners.

Thus, widespread, virulent, and archaic infections are greatly involved in natural selection for cognitive abilities. People infected with parasites may have brain damage and obvious maladaptive behavior in addition to visible signs of disease. Smarter people can more skillfully learn to distinguish safe non-polluted water and food from unsafe kinds and learn to distinguish mosquito infested areas from safe areas. Additionally, they can more skillfully find and develop safe food sources and living environments. Given this situation, preference for smarter child-bearing/rearing partners increases the chance that their descendants will inherit the best resistance alleles, not only for immune system resistance to disease, but also smarter brains for learning skills in avoiding disease and selecting nutritious food. When people search for mates based on their success, wealth, reputation, disease-free body appearance, or psychological traits such as benevolence or confidence; the effect is to select for superior intelligence that results in superior disease resistance.

Ecological dominance-social competition model

Another model describing the evolution of human intelligence is ecological dominance-social competition (EDSC), explained by Mark V. Flinn, David C. Geary and Carol V. Ward based mainly on work by Richard D. Alexander. According to the model, human intelligence was able to evolve to significant levels because of the combination of increasing domination over habitat and increasing importance of social interactions. As a result, the primary selective pressure for increasing human intelligence shifted from learning to master the natural world to competition for dominance among members or groups of its own species.

As advancement, survival and reproduction within an increasing complex social structure favored ever more advanced social skills, communication of concepts through increasingly complex language patterns ensued. Since competition had shifted bit by bit from controlling "nature" to influencing other humans, it became of relevance to outmaneuver other members of the group seeking leadership or acceptance, by means of more advanced social skills. A more social and communicative person would be more easily selected.

Intelligence dependent on brain size

Human intelligence is developed to an extreme level that is not necessarily adaptive in an evolutionary sense. Firstly, larger-headed babies are more difficult to give birth to and large brains are costly in terms of nutrient and oxygen requirements. Thus the direct adaptive benefit of human intelligence is questionable at least in modern societies, while it is difficult to study in prehistoric societies. Since 2005, scientists have been evaluating genomic data on gene variants thought to influence head size, and have found no evidence that those genes are under strong selective pressure in current human populations. The trait of head size has become generally fixed in modern human beings.

While decreased brain size has strong correlation with lower intelligence in humans, some modern humans have brain sizes as small as with Homo erectus but normal intelligence (based on IQ tests) for modern humans. Increased brain size in humans may allow for greater capacity for specialized expertise.

Expanded cortical regions

The two major perspectives on primate brain evolution are the concerted and mosaic approaches. In the concerted evolution approach, cortical expansions in the brain are considered to be a by-product of a larger brain, rather than adaptive potential. Studies have supported the concerted evolution model by finding cortical expansions between macaques and marmosets are comparable to that of humans and macaques. Researchers attribute this result to the constraints on the evolutionary process of increasing brain size. In the mosaic approach, cortical expansions are attributed to their adaptive advantage for the species. Researchers have attributed hominin evolution to mosaic evolution.

Simian primate brain evolution studies show that specific cortical regions associated with high-level cognition have demonstrated the greatest expansion over primate brain evolution. Sensory and motor regions have showcased limited growth. Three regions associated with complex cognition include the frontal lobe, temporal lobe, and the medial wall of the cortex. Studies demonstrate that the enlargement in these regions is disproportionately centered in the temporoparietal junction (TPJ), lateral prefrontal cortex (LPFC), and anterior cingulate cortex (ACC). The TPJ is located in the parietal lobe and is associated with morality, theory of mind, and spatial awareness. Additionally, the Wernicke's area is located in the TPJ. Studies have suggested that the region assists in language production, as well as language processing. The LPFC is commonly associated with planning and working memory functions. The Broca's area, the second major region associated with language processing, is also located in the LPFC. The ACC is associated with detecting errors, monitoring conflict, motor control, and emotion. Specifically, researchers have found that the ACC in humans is disproportionately expanded when compared to the ACC in macaques.

Fossils show that although Homo sapiens' total brain volume approached modern levels as early as 300,000 years ago, parietal lobes and cerebella grew relative to total volume after this point, reaching current levels of variation at some point between the approximate dates of 100,000 and 35,000 years ago.

Studies on cortical expansions in the brain have been used to examine the evolutionary basis of neurological disorders, such as Alzheimer's disease. For example, researchers associate the expanded TPJ region with Alzheimer's disease. However, other researchers found no correlation between expanded cortical regions in the human brain and the development of Alzheimer's disease.

Cellular, genetic, and circuitry changes

Human brain evolution involves cellular, genetic, and circuitry changes. On a genetic level, humans have a modified FOXP2 gene, which is associated with speech and language development. The human variant of the gene SRGAP2, SRGAP2C, enables greater dendritic spine density which fosters greater neural connections. On a cellular level, studies demonstrate von Economo neurons (VENs) are more prevalent in humans than other primates. Studies show that VENs are associated with empathy, social awareness and self-control. Studies show that the striatum plays a role in understanding reward and pair-bond formation. On a circuitry level, humans exhibit a more complex mirror neuron system, greater connection between the two major language processing areas (Wernicke's area and Broca's area), and a vocal control circuit that connects the motor cortex and brain stem. The mirror neuron system is associated with social cognition, theory of mind, and empathy. Studies have demonstrated the presence of the mirror neuron system in both macaques in humans; However, the mirror neuron system is only activated in macaques when observing transitive movements.

Group selection

Group selection theory contends that organism characteristics that provide benefits to a group (clan, tribe, or larger population) can evolve despite individual disadvantages such as those cited above. The group benefits of intelligence (including language, the ability to communicate between individuals, the ability to teach others, and other cooperative aspects) have apparent utility in increasing the survival potential of a group.

In addition, the theory of group selection is inherently tied to Darwin's theory of natural selection. Specifically, that "group-related adaptations must be attributed to the natural selection of alternative groups of individuals and that the natural selection of alternative alleles within populations will be opposed to this development".

Between-group selection can be used to explain the changes and adaptations that arise within a group of individuals. Group-related adaptations and changes are a byproduct of between-group selection as traits or characteristics that prove to be advantageous in relation to another group will become increasingly popular and disseminated within a group. In the end, increasing its overall chance of surviving a competing group.

However, this explanation cannot be applied to humans (and other species, predominantly other mammals) that live in stable, established social groupings. This is because of the social intelligence that functioning within these groups requires from the individual. Humans, while they are not the only ones, possess the cognitive and mental capacity to form systems of personal relationships and ties that extend well beyond those of the nucleus of family. The continuous process of creating, interacting, and adjusting to other individuals is a key component of many species' ecology.

These concepts can be tied to the social brain hypothesis, mentioned above. This hypothesis posits that human cognitive complexity arose as a result of the higher level of social complexity required from living in enlarged groups. These bigger groups entail a greater amount of social relations and interactions thus leading to an expanded quantity of intelligence in humans. However, this hypothesis has been under academic scrutiny in recent years and has been largely disproven. In fact, the size of a species' brain can be much better predicted by diet instead of measures of sociality as noted by the study conducted by DeCasien et al. They found that ecological factors (such as: folivory/frugivory, environment) explain a primate brain size much better than social factors (such as: group size, mating system).

Nutritional status

Early hominins dating back to pre 3.5 Ma in Africa ate primarily plant foods supplemented by insects and scavenged meat. Their diets are evidenced by their 'robust' dento-facial features of small canines, large molars, and enlarged masticatory muscles that allowed them to chew through tough plant fibers. Intelligence played a role in the acquisition of food, through the use of tool technology such as stone anvils and hammers.

There is no direct evidence of the role of nutrition in the evolution of intelligence dating back to Homo erectus, contrary to dominant narratives in paleontology that link meat-eating to the appearance of modern human features such as a larger brain. However, scientists suggest that nutrition did play an important role, such as the consumption of a diverse diet including plant foods and new technologies for cooking and processing food such as fire.

Diets deficient in iron, zinc, protein, iodine, B vitamins, omega 3 fatty acids, magnesium and other nutrients can result in lower intelligence either in the mother during pregnancy or in the child during development. While these inputs did not have an effect on the evolution of intelligence they do govern its expression. A higher intelligence could be a signal that an individual comes from and lives in a physical and social environment where nutrition levels are high, whereas a lower intelligence could imply a child, its mother, or both, come from a physical and social environment where nutritional levels are low. Previc emphasizes the contribution of nutritional factors to elevations of dopaminergic activity in the brain, which may have been responsible for the evolution of human intelligence since dopamine is crucial to working memory, cognitive shifting, abstract, distant concepts, and other hallmarks of advanced intelligence.