Intervention science has set for itself a noble goal. How do we reduce mental health problems, promote happiness, and help people to engage in behaviour that is effective and in their best interest? The scientific community has now spent hundreds of millions of dollars and decades to answer this question. The good news is that we have made excellent progress. Meta-analyses suggest that a wide variety of interventions are effective in reducing mental illness¹, increasing well-being², and promoting effective health³ and work behavior⁴.

Despite this success, Hayes and Hofmann⁵ argue that the dominant approach to intervention research may no longer be adequate. Meta-analytic research supports their view. Psychotherapy effect sizes are modest (about .30), when compared to placebo or treatment as usual⁶. Perhaps most concerning, effect sizes appear to have stagnated⁷. The authors argue that the lack of progress is not due to a lack of effort. Rather, they identify some major problems with the “protocol-for-disease” research paradigm, which seeks to identify effective clinical protocols to treat latent diseases.

Decades of research have failed to identify psychological diseases that exist independently of their so-called symptoms. We diagnose depression in individuals because they report feeling extremely sad and inactive, and then we say that they are inactive because they are depressed. In medicine, the physical disease can exist independent­ly of symptoms: someone can have cancer with or without symptoms of fatigue and nausea. If we abandon the assumption that a latent disease causes depression, we can free practitioners from the medical model and all its assumptions about suffering being caused by some internal abnormality.

We can open up to the role of context, and see that people display patterns of de­pressive symptoms that are causally relat­ed in different ways. For example, two clients have both received a diagnosis of depression. With only this knowledge, the practitioner might apply the same treatment protocol to both of them. What if we assume that they do not have the same disease? Instead, we look at the pattern of symptoms and how they interrelate in context. Imagine we discover that one of the depressed clients has just lost her partner, which is leading to intense sadness that drives social withdrawal, while the other client has been bullied at work, leading to social anxiety, which drives social withdrawal and intense sadness. Though we diagnose both clients with depression, we will presumably not give them the same intervention.

The protocol-for-disease approach does not recognize the role of contextual factors in therapeutic outcome⁵. Therapeutic procedures are not effective across all people and contexts. Some clients may love structured mindfulness practice, whereas others find such practices anxiety provoking and decidedly unhelpful⁸. Moreover, the protocol-for-disease approach focuses excessively on trademarked packages rather than evidence-based processes. It also fails to recognize the common effective processes shared by different protocols. A protocol is not a single thing, like a 50 mg dose of penicillin. Some processes are useful to a particular individual, some useless.

Hayes and Hofmann propose a radically new way forward, which, if correct, would lead to a revolution in intervention science. Rather than focusing on protocols for dis­eases, they focus on individualized processes of change for promoting broad and flexible behavioural repertoires. Their unifying framework allows people from any therapeutic approach to share a common process language focused on cognition, affect, attention, self, motivation, and overt behaviour.

Importantly, the framework shows how to tailor interventions for a particular person, in a particular context. Rather than assuming that a process, say emotional openness, has the same beneficial effect on everybody, it seeks to identify how different processes function, or drive well-being for different people. The practitioner identifies, through functional analysis, what processes are helping the client, and what processes are inert and harmful, and emphasizes the effective processes. This means that some aspects of an evidence-based protocol may be discarded, at least for a particular client.

Hayes and Hofmann are trying to entirely change the rules of the game. Shifting to their new process paradigm will not be easy. Improvements will not be immediate, just as the shift from Ptolemaic to Copernican system did not immediately result in better predictions⁹. We should expect null results and missteps along the way. Making matters worse, the current academic environment is not conducive to revolution. Academia pressures scientists to publish fast and efficiently in the top journals, and this usually means staying within accepted and safe paradigms, such as evaluating protocols for hypothesized latent diseases. The alternative path is uncertain and could be inefficient, at least initially. Yet it may lead to something new and potentially exciting.

The scientific community must decide whether to spend 20 more years showing that standardized protocols perform better than placebo, but not better than other protocols. Or to take risks, make some mistakes, and see if it can create personalized interventions that help each individual reach his/her full potential.

干预科学为自己设定了一个崇高的目标。我们如何减少心理健康问题，促进幸福，并帮助人们采取有效且符合他们最大利益的行为？科学界现在已经花费了数亿美元和数十年的时间来回答这个问题。好消息是我们已经取得了巨大的进展。荟萃分析表明，多种干预措施可有效减少精神疾病¹、增加幸福感²以及促进有效的健康³和工作行为⁴。

尽管取得了这一成功，海耶斯和霍夫曼⁵认为干预研究的主导方法可能不再足够。荟萃分析研究支持了他们的观点。与安慰剂或常规治疗相比，心理治疗的效果较小（约 0.30）⁶。也许最令人担忧的是，效应大小似乎停滞不前⁷。作者认为，缺乏进展并不是因为缺乏努力。相反，他们发现了“疾病方案”研究范式的一些主要问题，该范式旨在确定治疗潜在疾病的有效临床方案。

数十年的研究未能识别出独立于所谓症状而存在的心理疾病。我们诊断个体抑郁症是因为他们报告感到极度悲伤和不活跃，然后我们说他们不活跃是因为他们抑郁。在医学上，身体疾病可以独立于症状而存在：某人可能患有癌症，伴有或不伴有疲劳和恶心症状。如果我们放弃潜在疾病导致抑郁症的假设，我们就可以将医生从医学模型及其所有关于痛苦是由某些内部异常引起的假设中解放出来。

我们可以开放背景的作用，并看到人们表现出的抑郁症状模式以不同的方式存在因果关系。例如，两名客户都被诊断出患有抑郁症。仅凭这些知识，医生就可以对他们两人应用相同的治疗方案。如果我们假设他们没有相同的疾病怎么办？相反，我们着眼于症状的模式以及它们在上下文中如何相互关联。想象一下，我们发现一位抑郁的客户刚刚失去了她的伴侣，这会导致强烈的悲伤，从而导致社交退缩，而另一位客户在工作中受到欺凌，导致社交焦虑，从而导致社交退缩和强烈的悲伤。尽管我们诊断出两位客户都患有抑郁症，但我们可能不会给予他们相同的干预措施。

针对疾病的方案方法没有认识到背景因素在治疗结果中的作用⁵。治疗程序并非对所有人群和情况都有效。有些客户可能喜欢结构化的正念练习，而另一些客户则发现这种练习会引起焦虑并且毫无帮助⁸。此外，针对疾病的方案方法过度关注商标包装，而不是基于证据的流程。它也无法识别不同协议共享的共同有效流程。方案并不是单一的东西，就像 50 毫克剂量的青霉素一样。有些过程对特定个人有用，有些则无用。

海耶斯和霍夫曼提出了一种全新的前进方式，如果正确的话，将引发干预科学的一场革命。他们不关注疾病的治疗方案，而是关注个性化的改变过程，以促进广泛而灵活的行为方式。他们的统一框架允许来自任何治疗方法的人们共享一种共同的过程语言，重点关注认知、情感、注意力、自我、动机和外在行为。

重要的是，该框架展示了如何在特定情况下为特定人量身定制干预措施。它不是假设一个过程，比如情感开放，对每个人都有同样的有益影响，而是试图确定不同的过程如何发挥作用，或者为不同的人带来福祉。从业者通过功能分析来识别哪些流程对客户有帮助，哪些流程是惰性和有害的，并强调有效的流程。这意味着基于证据的协议的某些方面可能会被丢弃，至少对于特定的客户而言。

海耶斯和霍夫曼正试图彻底改变游戏规则。转向新的流程范式并不容易。改进不会立竿见影，就像从托勒密体系到哥白尼体系的转变并没有立即带来更好的预测一样⁹。我们应该预料到一路上会出现无效结果和失误。更糟糕的是，目前的学术环境不利于革命。学术界迫使科学家在顶级期刊上快速有效地发表文章，这通常意味着要遵守公认的安全范式，例如评估假设的潜在疾病的方案。替代路径是不确定的，并且可能效率低下，至少在最初是这样。然而，它可能会带来一些新的、令人兴奋的东西。

科学界必须决定是否再花 20 年时间来证明标准化方案的效果优于安慰剂，但并不比其他方案更好。或者冒险，犯一些错误，看看是否可以制定个性化的干预措施，帮助每个人充分发挥其潜力。