Augmented Humanity

This is not a test of an intervention. The book is a conceptual analysis and synthesis of existing theories from cognitive science, behavioral economics, organizational psychology, and digital humanities. It examines:

**The concept of "augmented agency"** — how digitalization expands intelligent processing capability beyond the individual human mind.

**Bounded rationality in digital contexts** — how algorithms and digital tools alter the classic limits on human rationality (time, information, cognitive capacity).

**Novel biases and myopias** — new cognitive distortions that emerge specifically when humans interact with digital systems (e.g., automation bias, algorithmic aversion, filter bubbles).

**Cognitive intersubjectivity** — how shared digital environments change how we understand each other's minds and intentions.

**Learning from performance** — how digital feedback loops (e.g., A/B testing, real-time analytics) change how individuals and organizations learn.

**Agentic self-generation** — how digital tools allow people to create and modify their own identities, capabilities, and decision-making environments.

The "comparator" is implicitly: human cognition *without* digital augmentation vs. human cognition *with* digital augmentation. But no experiment is run.

No human subjects were studied. The book is a theoretical work drawing on:

Published empirical studies from cognitive science, behavioral economics, and organizational psychology (not systematically reviewed — this is not a meta-analysis).

Philosophical and sociological theories of agency, rationality, and technology.

Case examples from business, finance, and digital culture (anecdotal, not systematically collected).

The "sample" is the author's selection of literature and theoretical frameworks. There is no systematic search strategy, no inclusion/exclusion criteria, and no quantitative synthesis.

No measurements were taken. The book uses conceptual analysis:

**Theoretical constructs** are defined and compared (e.g., "bounded rationality" vs. "digitally augmented rationality").

**Mechanisms** are proposed (e.g., "digital tools reduce search costs but introduce new confirmation biases").

**Effects** are hypothesized (e.g., "augmented agency may increase decision speed but decrease depth of understanding").

There are no instruments, scales, or quantitative metrics. The book does not report any effect sizes, confidence intervals, or p-values.

### Study design

This is a **theoretical monograph** — a book-length argument that synthesizes existing ideas and proposes a new conceptual framework. It is not:

A randomized controlled trial (RCT)

A systematic review or meta-analysis

A cohort study

A qualitative empirical study (e.g., interviews, ethnography)

A computational simulation

### What the design can and cannot prove

**What it can do:**

Identify gaps in existing research and propose new research questions.

Offer a coherent vocabulary and framework for thinking about human-digital interaction.

Generate hypotheses that could be tested in future empirical work.

**What it cannot do:**

Establish causal relationships (e.g., "digital augmentation causes X change in decision-making").

Quantify effect sizes or prevalence of any phenomenon.

Provide evidence that any proposed mechanism actually operates in real-world settings.

Compare the effectiveness of different digital augmentation strategies.

### Methodological weaknesses

**No systematic literature search** — the author selects theories and studies that support the argument; there is no transparent method for including or excluding evidence.

**No empirical data** — all claims are theoretical or based on cherry-picked examples.

**No falsifiable predictions** — the framework is so broad that it can explain almost any outcome post-hoc.

**No peer review of the specific claims** — while the book may have been reviewed as a whole, individual claims are not vetted through the standard scientific process.

**No replication** — even if the framework were testable, no replication attempts are reported.

Since this is not an empirical study, there are no findings in the scientific sense. The book's main **theoretical propositions** include:

**Digitalization expands the "intelligent processing capability" available to humans** — but this expansion is not simply additive; it creates new forms of interdependence between human and machine cognition.

**Bounded rationality is altered, not eliminated** — digital tools reduce some cognitive limits (e.g., memory, calculation speed) but introduce new ones (e.g., over-reliance on algorithmic outputs, reduced ability to reason without digital aids).

**Novel biases emerge in human-AI interaction** — including automation bias (trusting machine outputs too much), algorithmic aversion (distrusting algorithms after seeing them err), and filter bubble effects (reinforcing existing beliefs through personalized content).

**Cognitive intersubjectivity is mediated by digital platforms** — shared digital environments change how people infer each other's mental states, potentially reducing empathy or creating new forms of coordination.

**Learning from performance becomes faster but more shallow** — real-time feedback loops (e.g., A/B testing, dashboards) allow rapid adjustment but may discourage deep understanding of underlying causal mechanisms.

**Agentic self-generation becomes more fluid** — people can use digital tools to create multiple identities, automate parts of their decision-making, and redesign their own cognitive environments.

**No quantitative results are reported.** No effect sizes, no confidence intervals, no p-values.

Not applicable. The book does not provide any quantitative estimates of effect magnitude. For example, it does not claim "people with access to AI decision support make decisions 23% faster with 12% less accuracy" — it only argues that such trade-offs exist in principle.

### What the author acknowledges (implicitly)

The book is a "mapping of terrain" for a future science, not a completed science.

The mechanisms described are "novel" and require empirical testing.

The framework is cross-disciplinary and may not satisfy specialists in any one field.

### What a critical reader would note

**No empirical evidence** — every claim in the book is a hypothesis, not a finding. A self-experimenter cannot take any of these claims as established fact.

**Selection bias** — the author chooses examples and theories that support the argument. Counter-evidence (e.g., studies showing digital tools improve deep reasoning, or cases where human-AI collaboration fails catastrophically) is not systematically addressed.

**No operational definitions** — key terms like "augmented agency," "intelligent processing capability," and "cognitive intersubjectivity" are defined conceptually but not in a way that allows measurement or testing.

**No practical guidance** — the book does not tell a reader how to augment their own cognition, what tools to use, or what outcomes to expect.

**Publication context** — the book is published as an open-access academic monograph, but the journal/publisher is listed as "Unknown" and the year is 2021. Without peer-reviewed journal publication, the claims have not been vetted by domain experts.

**Scope creep** — the book attempts to cover everything from individual cognition to organizational behavior to digital culture, which makes it difficult to derive specific, testable predictions.

### For someone running their own n=1 experiment

**Important caveat:** This book provides no empirical results to base an experiment on. The following takeaways are derived from the *hypotheses* the book proposes — they are suggestions for what you *could* test, not what the book has proven.

### What to test (specific intervention and dose)

**Hypothesis 1: AI-assisted decision-making** — Use a large language model (e.g., ChatGPT, Claude) to generate options or evaluate trade-offs before making a decision. Compare to making the same type of decision without AI assistance.

**Hypothesis 2: Digital feedback loops** — Use a habit-tracking app with real-time feedback (e.g., streaks, notifications) vs. a simple paper diary for tracking a behavior (e.g., daily exercise, meditation).

**Hypothesis 3: Reduced search costs** — Use a search tool that aggregates information (e.g., Perplexity, Google Scholar) vs. manual searching through individual sources for a research question.

**Dose:** For AI assistance, try using it for 5–10 decisions per day for 2 weeks. For feedback loops, use the app for 30 days. For search, use the tool for 1 hour per day for 1 week.

### Minimum meaningful duration

**For decision-making experiments:** At least 2 weeks per condition to get enough decisions for statistical power. A crossover design (2 weeks with AI, 2 weeks without, then repeat) would be stronger.

**For feedback loop experiments:** At least 30 days to observe habit formation effects. The book suggests that digital feedback may produce faster but shallower learning — so you need enough time to see if the effect plateaus or reverses.

**For search experiments:** At least 1 week per condition, with multiple search tasks per day.

### What to measure (specific metrics)

**Decision quality:** For decisions with objectively correct answers (e.g., factual questions, puzzles), measure accuracy (percentage correct). For subjective decisions (e.g., which job offer to take), measure satisfaction on a 1–10 scale 1 week and 1 month after the decision.

**Decision speed:** Time from start of deliberation to final choice (in minutes).

**Cognitive effort:** Subjective rating of mental effort (1–10 scale) after each decision. Or use a simple reaction time test before and after each decision session to measure cognitive fatigue.

**Depth of understanding:** After each decision, ask yourself to explain the reasoning behind your choice in 2–3 sentences. Rate your confidence in that explanation (1–10). Have a blind rater (or an AI) rate the quality of the explanation.

**Bias measures:** After each decision, note whether you felt over-reliant on the digital tool (automation bias) or unduly skeptical of it (algorithmic aversion). Use a 1–5 scale.

### Key confounds to control for

**Order effects:** If you test AI-assisted first, you may get better simply from practice. Use a crossover design (A-B-A-B) or randomize the order of conditions.

**Expectation effects:** If you believe AI will help, you may try harder or rate outcomes more favorably. Use blinding where possible (e.g., have a friend set up the AI tool so you don't know if it's active).

**Task difficulty:** If the tasks are too easy, AI won't help. If they're too hard, AI might hurt. Calibrate tasks to be moderately challenging (70–80% accuracy without AI).

**Time of day:** Decision quality varies with circadian rhythms. Do all decision tasks at the same time of day.

**Sleep and stress:** Track sleep quality (e.g., with a sleep diary) and daily stress (1–10 scale) as covariates.

**Tool familiarity:** If you're new to the AI tool, you may get worse before you get better. Run a 3-day practice period before collecting data.

### What a positive result would look like

**For AI-assisted decisions:** You see a consistent improvement in accuracy (e.g., +10–15 percentage points) with no increase in decision time, or a reduction in decision time (e.g., 20–30% faster) with no loss of accuracy. You also see no increase in automation bias (i.e., you don't blindly trust the AI when it's wrong).

**For digital feedback loops:** You see faster habit acquisition (e.g., reaching 80% adherence in 14 days vs. 21 days with paper diary) but also a faster drop-off when the feedback stops (e.g., adherence drops to 50% within 7 days vs. 14 days for paper).

**For reduced search costs:** You find relevant information faster (e.g., 15 minutes vs. 30 minutes per search task) but your ability to recall the information from memory 1 week later is worse (e.g., 40% recall vs. 60% recall).

**What a null result would look like:** No difference between conditions on any metric, or a difference that is inconsistent across tasks (e.g., AI helps on factual questions but hurts on creative ones). This would suggest that the benefits of digital augmentation are task-specific and not universal — which is exactly the kind of nuance the book's framework would predict.

**What a negative result would look like:** Digital augmentation consistently makes you slower, less accurate, or less satisfied. This would challenge the book's optimistic framing and suggest that the "novel myopias and biases" dominate the benefits for your specific cognitive style or task domain.

### Summary for the self-experimenter

This book is a **theoretical map**, not a **user manual**. It can help you generate hypotheses about how digital tools might change your cognition, but it provides no data on what actually works. If you want to run an experiment based on this book, you must design it yourself, measure everything carefully, and be prepared for null or negative results. The most valuable contribution of the book is its list of potential mechanisms (automation bias, shallow learning, altered rationality) — these give you specific things to measure and watch for in your own behavior.