Tekinsiz Valley

Uncanny Valley (English: Uncanny Valley) is a hypothesis that describes how, as the appearance and behavior of a human-like entity such as a robot or animation become increasingly human-like, the emotional response of an observer toward it initially becomes more positive—but at a certain threshold where full human realism is not quite achieved, it suddenly shifts to negative feelings such as eeriness, fear, or revulsion. When human likeness approaches perfection, however, this negative response disappears and is replaced once again by positive emotions. This abrupt dip in the emotional response curve is named a “valley” due to its resemblance to a valley on a graphical plot.

Human-Like But Not Quite (Generated by Artificial Intelligence)

Definition and Historical Development

The concept of the Uncanny Valley was first introduced in 1970 by robotics professor Masahiro Mori in a Japanese journal called Energy, in an article titled “Bukimi no Tani Genshō.” In this paper, Mori predicted that human emotional responses to human-like robots do not increase linearly with increasing similarity; rather, they experience a sharp decline at a specific threshold. Although the article did not gain widespread attention upon publication, it later became one of the foundational references in fields such as robotics, computer graphics, and psychology.

Mori’s notion of “uncanniness” also parallels earlier psychoanalytic and philosophical ideas. The concept’s origins can be traced back to a 1906 article by German physician Ernst Anton Jentsch titled “On the Psychology of the Uncanny.” In this work, Jentsch argued that the unsettling effect produced by figures such as wax dolls or automatons stems from a cognitive uncertainty about whether they are alive or inanimate.

Later, Sigmund Freud addressed the topic in his 1919 essay “Das Unheimliche” (“The Uncanny”), analyzing it from aesthetic and psychological perspectives. According to Freud, the uncanny arises when something familiar and previously known reappears in a strange and unsettling form after being repressed. Freud described this experience as occurring when the boundary between imagination and reality blurs, noting that this feeling finds a more fertile ground in fictional characters than in real life. Building on these ideas, Mori transformed the concept into a concrete hypothesis centered on robots and human-like objects.

Uncanny Valley Graph and Core Concepts

The Uncanny Valley hypothesis is typically illustrated using a two-axis graph. The horizontal axis (x-axis) represents the level of human likeness, while the vertical axis (y-axis) represents the observer’s perceived level of affinity or closeness (shinwakan).

• Initial Rise (First Peak): Objects with low human likeness, such as an industrial robotic arm, do not evoke a strong sense of affinity. However, as the object acquires basic human features such as a face, arms, and legs—for example, a toy robot—affinity and sympathy toward it increase. This forms the first peak on the graph.

• Descent into the Valley: As human likeness increases further but still falls short of full realism, an abrupt shift in response is observed. For instance, a highly realistic prosthetic hand may appear human at first glance. Yet, upon touch, its coldness, lifelessness, and texture differences trigger a shiver and a sense of uncanniness in the observer. This phenomenon pushes the object to the lowest point of the valley. Similarly, corpses or zombie figures also reside deep within this valley.

• Emergence from the Valley (Second Peak): When human likeness continues to increase until the object becomes nearly indistinguishable from a healthy human, the feeling of uncanniness vanishes and the observer’s response returns to a highly positive level. This constitutes the second and highest peak on the graph.

The Effect of Movement

Mori noted that movement intensifies the Uncanny Valley effect. A stationary industrial robot is perceived merely as a machine part, but when it begins to move in a human-like manner, it may elicit greater sympathy. Conversely, if a prosthetic hand already situated at the bottom of the valley begins to move, the sense of uncanniness becomes even more pronounced. Since movement is interpreted as a sign of life, an entity that moves like a human but is known not to be human becomes more disturbing due to this contradiction.

Theoretical Approaches and Explanations

Various scientific hypotheses have been developed to explain the Uncanny Valley phenomenon. These explanations can generally be grouped into two main categories: evolutionary psychology and cognitive dissonance.

Evolutionary Psychology-Based Explanations

These approaches argue that the feeling of uncanniness stems from human survival and reproductive instincts.

• Threat Avoidance Hypothesis: This hypothesis suggests that uncanniness functions as a defensive mechanism against potential threats.

• Pathogen Avoidance: Imperfections in human-like entities—such as abnormal skin tone or asymmetrical facial features—subconsciously evoke associations with illness, genetic disorder, or death. This triggers an evolved aversion response aimed at avoiding potential carriers of disease.

• Mortality Salience: Realistic yet lifeless figures remind individuals of their own mortality and corpses. According to terror management theory, this acts as a factor that triggers death anxiety.

• Evolutionary Aesthetics Hypothesis: According to this view, uncanniness fundamentally arises from aversion toward individuals who lack aesthetically attractive traits. Humans have developed preferences for aesthetic features linked to health and fertility—such as symmetry and smooth skin. Entities in the Uncanny Valley fail to meet these aesthetic standards, resulting in negative reactions.

Cognitive Dissonance-Based Explanations

These approaches assert that the feeling of uncanniness arises from mental contradictions and violated expectations.

Mind Perception Hypothesis

This hypothesis suggests that human-like robots prompt observers to attribute mental capacities such as feeling and perception to them. However, since these capacities are assumed to be uniquely human, the idea that they could exist in a machine creates discomfort.

Violation of Expectation Hypothesis

A human-like appearance generates expectations that the entity will behave like a human—speaking naturally, moving fluidly, and responding emotionally. When the robot’s mechanical movements or synthetic voice fail to meet these expectations, a mismatch between appearance and behavior produces the sense of uncanniness.

Categorical Uncertainty Hypothesis

This hypothesis argues that uncanniness stems from difficulty in categorizing an entity as either “human” or “machine.” This ambiguity creates cognitive discomfort. However, empirical studies testing this hypothesis have yielded inconsistent results.

Applications and Related Disciplines

The Uncanny Valley hypothesis influences design and interaction strategies in numerous fields, particularly robotics and computer graphics.

Robotics and Design

Robot designers typically adopt one of two strategies to avoid the uncanny effect. The first is to deliberately design entities that are clearly non-human, resembling cartoon characters, thereby remaining on the first peak of the graph. The second is to push technological boundaries to create designs so realistically human-like that they surpass the valley entirely and appear as convincing as a healthy human. Mori suggested, drawing from the example of eyeglasses which are not visually identical to human eyes yet create appealing new forms, that elegant and functional designs that avoid excessive realism may offer a safer level of affinity.

Computer Graphics and Animation (CGI)

This hypothesis is particularly considered in the design of computer-generated characters used in films and video games. Some animation studios, such as Pixar, have avoided high levels of photorealism in favor of more stylized characters, illustrating this strategy.

Artificial Intelligence Influencers (AI Influencers)

Today, AI influencers on social media—designed using CGI—are examined within the framework of the Uncanny Valley theory. Their extreme realism theoretically carries a risk of being perceived as uncanny. However, some research indicates that these virtual characters receive predominantly positive responses (attractive, successful, appealing) from their followers, suggesting that the boundaries of the Uncanny Valley may be overcome or altered.

Social and Philosophical Debates

The concept is also used metaphorically to question our relationship with technology. As technology—particularly digitalization and artificial intelligence—penetrates every aspect of life, concerns about control and privacy arise. This situation is sometimes interpreted as society entering a kind of “uncanny valley.”

Terminology and Methodological Limitations

Scientific research on the Uncanny Valley faces certain terminological and methodological challenges.

Terminology

The Japanese term “shinwakan” used by Mori in his original text was initially translated into English as “familiarity,” but later corrected to “affinity.” This inconsistency has led to ambiguities in defining and measuring the concept. The use of other terms such as “likability” and “attractiveness” further complicates the comparison of findings across studies.

Evaluation Methods

Most studies rely on surveys and Likert-scale questions measuring participants’ subjective emotional responses. However, such self-reported assessments can be influenced by factors like social desirability bias. In recent years, more objective physiological measurement methods—including fMRI (functional magnetic resonance imaging), EMG (facial electromyography), and eye tracking—have been increasingly adopted to address these limitations.

Stimulus Selection

The lack of standardization in stimuli used in studies—such as images, videos, or robots—constitutes another factor affecting result consistency. Visuals collected from disparate sources or artificially generated using morphing techniques may introduce unwanted variables, reducing the reliability of findings.